RDP 2022-09: Estimating the Effects of Monetary Policy in Australia Using Sign-restricted Structural Vector Autoregressions 6. Conclusion

To estimate the effects of monetary policy in Australia, this paper imposes sign restrictions on impulse responses to a monetary policy shock, the monetary policy reaction function, and the relationship between a proxy for the monetary policy shock and the shock itself. The advantage of using these sign restrictions over traditional zero restrictions is that the sign restrictions are less restrictive and arguably less controversial, in the sense that they are consistent with a wide range of conventional macroeconomic models and/or prior information. The approach to inference that I take explicitly addresses the problem of posterior sensitivity to the choice of prior that arises when conducting Bayesian inference under sign restrictions, which turns out to be important.

Together, the sign restrictions that I consider appear to be reasonably informative about the effects of a standard deviation monetary policy shock on output and prices; output and prices fall with high posterior probability at horizons beyond a year or so, and this result is robust to the choice of prior. However, the restrictions appear to be quite uninformative about responses to a 100 basis point monetary policy shock, which are the responses that are more relevant from a policymaking perspective, reflecting the fact that they do not necessarily rule out extremely small (or zero) responses of the cash rate to a monetary policy shock. Nevertheless, it remains possible to draw some useful inferences about the effects of a 100 basis point shock. For example, in response to a 100 basis point shock, there is strong evidence that output declines by at least half a per cent at the two-year horizon. The results are consistent with the output effects of a change in the cash rate lying towards the upper end of the range of existing estimates and, in particular, the results suggest that the response of output is likely to exceed the peak responses in the RBA's multi-sector and MARTIN models. One caveat around these results is that they are sensitive to shortening the estimation sample to the inflation-targeting period.

The results in this paper could be further refined by imposing additional credible identifying restrictions. For example, restrictions on the sign of the monetary policy shock or its relative contribution to changes in the cash rate in specific historical episodes could sharpen identification (e.g. Antolín-Díaz and Rubio-Ramírez 2018; Giacomini, Kitagawa and Read 2021a). Another potentially useful avenue would be to identify other structural shocks, which should in principle help to identify monetary policy shocks. The availability of additional proxies for Australian macroeconomic shocks would be useful in this regard. More generally, it is important that we continue to search for credible identifying restrictions and question the assumptions underlying existing estimates of the effects of monetary policy.