RDP 2010-07: Monetary Policy and the Exchange Rate: Evaluation of VAR Models 5. Estimated Sign-restricted VAR – Actual Data

Before concluding, we estimate a sign-restricted VAR with the set of sign restrictions given in Table 2^[14] and using the same Australian data used to estimate the DSGE model in Section 3. Figure 7 shows the impulse responses to a 25 basis point contractionary monetary policy shock. The effect of monetary policy on output and inflation is broadly in line with other Australian studies, which suggest that a shock of a similar magnitude reduces the level of output by around 0.2 percentage points relative to baseline within two years, and lowers the quarterly inflation rate by around 0.02 percentage points after two years (see for instance, Brischetto and Voss 1999, Dungey and Pagan 2000, Berkelmans 2005, Jääskelä and Nimark 2008 and Nimark 2009).

It can be seen that the range of responses surrounding the real exchange rate is dispersed, with around half of the responses indicating an appreciation on impact. This is not surprising; as we pointed out earlier, identifying too few shocks can make it difficult to recognise the qualitative and quantitative features of the true data generating process. Therefore, a richer set of sign restrictions and identified shocks than our simple model imposes might be required to capture the transmission mechanism.^[15] Using Australian data, Liu (forthcoming) estimates a slightly more complex sign-restricted VAR model that captures the movements in the terms of trade. His results, based on the central tendency measures, indicate that the response of the real exchange rate to a monetary policy shock is delayed, with a peak effect after about one year. However, we have also cautioned against the central tendency measures, showing in our controlled experiments that the true impulses hardly ever coincide with the median. It is therefore plausible in our case that the ‘true’ exchange rate response lies somewhere on the ‘lower’ tail of the empirical distribution, indicating a sizeable appreciation. Of course, if we believe strongly in the instantaneous appreciation of the exchange rate following a contractionary monetary policy shock, we could impose this restriction as in Farrant and Peersman (2006) and Bjϕrnland and Halvorsen (2008).

Footnotes

However, we impose sign restrictions on the interest rate and inflation for eight and two quarters, respectively. If sign restrictions are imposed for a shorter period of time, the interest rate becomes expansionary reasonably quickly. [14]

Another potential source of misspecification stems from non-stationarity of the data, see for instance, Dungey and Pagan (2009). [15]