RDP 2015-11: Unprecedented Changes in the Terms of Trade 6. Implications of Higher and More Volatile Commodity Prices

Our estimation sample ended in 2013:Q4. Since that time, commodity prices have fallen substantially. Are these developments consistent with the permanent increase in commodity prices that we estimate? To address this question, Figure 10 compares the posterior distribution of forecasts generated from our model to actual out-of-sample commodity price developments. As in Figure 6, the lines labelled ‘Long-run commodity prices’ show the posterior distribution of the estimated non-stochastic paths – that is, the path commodity prices would have taken in the absence of temporary shocks. In contrast, the lines labelled ‘Model forecasts’ show the posterior distributions of the model forecasts – that is, the path the model expected commodity prices to follow after 2013:Q4 along the transition to their long-run level, assuming no additional temporary shocks after the end of the sample. The forecasts and actual outcomes align closely. That is, recent commodity price developments are consistent with the idea that although a large part of the boom in commodity prices was temporary, a substantial portion was permanent.

Figure 10: Observed and Long-run Commodity Prices

The estimated changes in the unconditional mean and the volatility of the economy have implications for the structure of the economy as well as its sensitivity to commodity price shocks. Table 4 illustrates how the structure of the model economy changes as a result of the estimated 42 per cent increase in the long-run level of commodity prices. Other characteristics of the balanced growth path, including the growth rate of real quantities and rates of inflation are invariant in the long run to the level of commodity prices.

Table 4: Change in Economic Structure
Concept Initial structure Final structure
Expenditure (per cent of GDP)
Consumption 72.7 72.0
Investment 27.3 28.0
Exports 19.7 22.0
Production (per cent of GDP)
Non-tradeable 57.0 57.6
Home tradeable 36.2 32.3
Commodities 6.8 10.2
Exports (per cent of exports)
Resource exports 34.6 51.8
Other exports 65.4 48.2
Relative prices (index)
Real exchange rate 100.0 131.0
Note: The real exchange rate has been inverted so that an increase represents a real appreciation

The most striking consequence of the estimated increase in the long-run level of commodity prices is a shift in resources from the non-commodity tradeable sector to the commodity sector. The commodity sector's share of exports increases from 35 per cent to 52 per cent. And its share of value added increases from 6.8 per cent of GDP to 10.2 per cent. There is also a significant effect on the real exchange rate, which appreciates permanently by around 30 per cent.

Of course, it bears emphasising that these shifts in the model's steady state are estimates calculated at a single point in the parameter distribution. As Figure 7 illustrates, the standard errors surrounding some of these estimates are wide. Moreover, the shift in the model's steady state reflects only the effects of changes in the long-run level of commodity prices. In particular, it does not take into account factors, like unconventional monetary policy in many advanced economies, that have had a persistent influence on the level of Australia's real exchange rate in recent years.

The change in the economy's steady state and the change in the volatility of shocks to commodity prices alter the relative contribution that shocks have to the observable variables. To measure the implications of these changes, we compute unconditional variance decompositions at the estimated posterior mode for variables of interest in two regimes: a low commodity price and volatility regime (κ = 1 and σκ = .05) and a high commodity price and volatility regime (κ = 1.42 and σκ = .11) (Table 5).[18] We report only the contribution of commodity price shocks to the variances of these variables, as these show the greatest variation between the two regimes.

In the ‘low regime’, we estimate that shocks to commodity prices make a modest contribution to economic fluctuations. An exception is net exports, where shocks to commodity prices account for 15 per cent of the variance, and the level of the real exchange rate, where shocks to commodity prices account for about 3 per cent of the variance. In the ‘high regime’, the share of the variance of net exports and the level of the real exchange rate explained by shocks to commodity prices increases substantially. And commodity price shocks also make a modest, but noticeable, contribution to changes in the nominal exchange rate and non-tradeable inflation.

Table 5: Variance Decomposition
Per cent
Contribution to variance of Regime
Low High
Inflation 0.0 0.3
Non-tradeable inflation 0.5 3.7
GDP growth 0.1 1.2
Investment growth 0.1 1.3
Consumption growth 0.1 0.5
Nominal exchange rate growth 0.5 4.6
Real exchange rate 3.2 44.8
Net exports-to-GDP ratio 14.8 57.8
Nominal interest rate 0.2 2.4

We draw two implications from our work. First, in light of our estimates, and the behaviour of commodity prices since the end of our sample, we believe that commodity prices will remain higher than they were before the commodity price boom. In the near term, the fall in commodity prices from their peak is likely to diminish growth in economic activity. But the pain would have been far worse if Australian firms, households and policymakers had behaved as if all of the increase in commodity prices during the boom had been permanent.

Second, the sensitivity of Australian economic developments to commodity price movements during the peak of the commodities boom was unusual. In the future, commodity price developments will probably not exert as much influence over aggregate economic activity, consumption, or even investment. However, a permanently higher level of commodity prices and larger commodity price shocks mean that, relative to the period before the commodity price boom, the commodity sector will account for a larger share of the economy and commodity price movements will be more important, particularly for exports and the exchange rate.


We do not report variance decompositions for the high-κ low-σκ regime because the joint posterior implies a low probability for this regime. [18]