RDP 2013-09: Terms of Trade Shocks and Incomplete Information 5. What Shocks Have Occurred?

An advantage of estimating a structural model is that it sheds light on whether observed movements in the terms of trade have reflected changes in its transitory or persistent components. In this section, I discuss these estimates.

Figure 8 plots a series of estimates for the permanent component of the Australian terms of trade. The first shows the median estimate of g_t|T, derived by applying the Kalman smoother to the posterior distribution of the parameters. This is the model's estimate of the permanent component of terms of trade shocks, calculated using all information in the sample. In contrast, the second corresponds to the median estimate g_t|t. This is the model's estimate of agents' real time beliefs about the permanent component of terms of trade shocks. The shaded area represents the 95 per cent confidence interval for this value.

For the first part of the sample, the two lines move together fairly closely, although the real-time estimates appear to lag the full sample estimates slightly. That is, the model suggests that the beliefs that agents held about the permanent component of the terms of trade, formed given the information available to them at the time, were fairly close to what the model now suggests, given the full sample of data. The slight lag reflects the informational frictions in the model, which mean that agents generally only perceive changes in the long-run trend of the terms of trade after these changes have occurred. Nevertheless, agents' mistakes appear for the most part to have been fairly short-lived.

The pattern in the 2000s is different. Throughout this decade, the model suggests that agents' estimates of the permanent component of the terms of trade were systematically lower than what the model now suggests was the case. This is consistent with the terms of trade forecasts presented in Figure 2, and suggests that throughout this episode agents attributed more of the run-up in the terms of trade to its temporary component than was actually the case.

One can also use the Kalman smoother to derive estimates of the time series of structural shocks to the terms of trade. These are shown in Figure 9. Of particular interest is the sequence of large permanent shocks to the terms of trade in the 2000s. The model also estimates that there were large negative shocks to the permanent component of the terms of trade in the mid 1970s and late 1980s and, to a lesser extent, around the year 2000. It is also interesting to note that the model suggests that there were large transitory shocks to the terms of trade in 2007/08 – that is, at least part of the large increases in the terms of trade at this time were transitory. Moreover, the model attributes much of the recovery in the terms of trade after 2009 to a series of positive transitory shocks.