RDP 2015-07: A Multi-sector Model of the Australian Economy 7. Conclusion

This paper has outlined an estimated DSGE model of the Australian economy currently in use at the Reserve Bank of Australia. The model differs from other Australian DSGE models through the inclusion of multiple sectors, including non-tradeable, resource and non-resource tradeable sectors. We estimate the model using Bayesian methods over the inflation-targeting era. We then explore the consequences of shocks to monetary policy, exchange rates and resource prices and use the model to decompose the sources of Australian business cycle fluctuations over the past two decades.

Relative to previous models of the Australian economy, the multi-sector structure of our model has several benefits. Most importantly, it gives us a deeper understanding of how changes in interest rates, exchange rates and other macroeconomic variables affect the broader economy. Such an understanding is particularly important for small open economies, such as Australia, for which many shocks – for instance, resource price shocks – are sectoral in nature. Our estimation also highlighted important differences in the characteristics of the various sectors of the economy, such as the slope of their Phillips curves. An awareness of these sectoral differences can help us to better interpret the responses of aggregate variables to macroeconomic shocks.

The model can be used to provide scenario and sensitivity analysis. It also provides a crosscheck on forecasts produced by reduced-form econometric techniques and judgement. Like most macroeconomic models in use at policy institutions, this model is likely to evolve over time. Given the important role of dwelling investment in the transmission of monetary policy, the addition of housing as a non-durable consumption good is high on our research agenda. Another possible area for development is the inclusion of a more sophisticated labour market that features involuntary unemployment and a meaningful treatment of labour force participation. Incorporating a measure of interest rate spreads would also allow us to answer some interesting questions relating to financial markets. Finally, although the model has been designed primarily for scenario analysis, it may be interesting to examine its out-of-sample forecasting performance.