RDP 2015-07: A Multi-sector Model of the Australian Economy 4. The Dynamics of the Estimated Model

In this section, we explore some of the economic mechanisms at work in the estimated model by examining dynamic responses to unanticipated shocks to monetary policy, resource prices and risk premiums. This set of shocks is relevant for understanding the behaviour of the Australian economy over recent decades and allows us to compare our model to others in the literature. These are also shocks that illustrate how aggregate disturbances can have differing implications for the various sectors of the economy.

4.1 Monetary Policy Shocks

We start our discussion with a monetary policy shock, illustrated in Figures 2 (aggregate variables) and 3 (selected sectoral variables). We scale the shock so that the median cash rate response is 100 basis points on impact.^[10]

The domestic variables respond in a manner consistent with economic theory. Higher interest rates lead to a contraction in the real economy (relative to a baseline in which interest rates remained constant). Output follows a hump-shaped pattern, decreasing by 0.6 per cent at its trough, before returning to steady state after five years. The components of domestic demand behave in a similar manner. The response of consumption is around half as large as the response of GDP, while the contraction in investment is much larger. The nominal and real exchange rates both appreciate, which lowers the inflation rate of imported goods and services. In conjunction with the slowdown in economic activity, this causes a decrease in CPI inflation.

Import volumes also contract following a positive monetary policy shock, as the effects of higher interest rates on domestic demand overwhelm the substitution effects associated with the stronger exchange rate. The response of export volumes, however, is unexpected; they increase on impact, before decreasing below their baseline level after 10 quarters. The initial increase in exports – which is common to both resource and non-resource exports – is due to lower domestic wages and cheaper capital, both of which reduce firms' costs.

In response to tighter monetary policy, output contracts in the non-tradeable and non-resource tradeable sectors, reflecting the sensitivity of these sectors to domestic demand conditions. In contrast, the decrease in resource output is delayed and reflects a persistent contraction in investment, which lowers this sector's productive capacity.

Inflation decreases in both the traded and non-traded sectors. However, the fall in tradeables inflation is larger, due to the fact that prices are estimated to be more flexible in this sector. Investment contracts by similar amounts in all of the domestic production sectors.

Several other Australian macroeconomic models have examined the aggregate impacts of monetary policy shocks and it is informative to compare their estimates against ours. The response of economic activity to monetary policy shocks is larger in our model than in the the DSGE models of Buncic and Melecky (2008) and Jääskelä and Nimark (2011) and the structural VAR model of Dungey and Pagan (2009). Like Jääskelä and Nimark, our expenditure responses display a hump shape, with the peak contraction in activity occurring after one year. The response of inflation in our model is smaller than in Jääskelä and Nimark, although it appears to be in line with other studies. Like us, Jääskelä and Nimark report a fall in imports in response to a monetary policy tightening. However, unlike us, they report an initial decrease in export volumes.^[11]

4.2 Risk Premium Shocks

We turn next to the effects of a shock to the risk premium term in the uncovered interest rate parity condition, shown in Figures 4 and 5. Technically, this shock alters the price at which domestic residents can sell foreign currency-denominated bonds to foreigners – thus acting to change the risk premium attached to these bonds. In practice, this shock causes a change in the nominal exchange rate that is unrelated to domestic or overseas economic conditions. In this sense it is similar to a pure ‘exchange rate’ shock. We calibrate the size of the shock so that it causes an initial 10 per cent appreciation of the nominal exchange rate.^[12]

The appreciation of the nominal exchange rate lowers the price of overseas goods relative to domestically produced goods. This leads households and firms to substitute away from domestic goods towards imports. As a consequence, export volumes decrease and import volumes increase. Aggregate output contracts by around 0.3 per cent over the first two quarters, before returning to its steady state over the subsequent two years.

Despite the contraction in economic activity, the appreciation of the exchange rate expands domestic demand. Consumption and investment both increase in a hump-shaped pattern, with a peak after six to eight quarters. The expansion in demand reflects two factors. First, the appreciation raises the purchasing power of domestic residents; for a given level of domestic production, they can now afford to consume more imports. Second, monetary policy responds to the shock by lowering the cash rate. Lower interest rates have a positive impact on domestic demand.

The decrease in interest rates is partly due to the contraction in real economic activity. It also reflects the deflationary impacts of the exchange rate appreciation, which lowers annualised inflation by 0.4 per cent on impact. The decrease in inflation is largely due to a fall in the rate of imported and domestically produced tradeable inflation. In contrast, the decrease in non-tradeable inflation is modest.

The contractionary effect of the exchange rate appreciation is concentrated in the tradeable parts of the economy; output in the resource and non-resource tradeable sectors both decline. In contrast, the decline in non-tradeable production is much smaller, and turns positive after four quarters, reflecting stronger domestic demand.

The dynamics of investment in the domestic production sectors differs from that of output. Although production in the non-tradeable and non-resource tradeable sector contracts, investment in those sectors is estimated to increase following an exchange rate shock. This is because the decrease in the price of investment goods induces firms in this sector to replace labour with capital in production. Investment in the resource sector also increases, albeit with a lag.

Using a structural VAR model, Manalo, Perera and Rees (in progress) find quantitatively similar results to us. At a sectoral level, they conclude that output in trade-exposed industries, including mining, manufacturing and other business services, experiences the largest contractions in activity following an exchange rate appreciation.^[13] In contrast, output in the construction and goods distribution industries, which have relatively less trade exposure, initially expands after an exchange rate appreciation. These results are broadly consistent with our findings.

4.3 Resource Price Shocks

Our third set of impulse responses shows the dynamic effects of a positive innovation to resource prices, illustrated in Figures 6 and 7. We scale the shock so that it causes a temporary 10 per cent increase in resource prices (in foreign currency terms).^[14]

Higher resource prices raise domestic income, causing a sustained expansion in domestic demand. Investment follows a hump-shape profile, peaking after six quarters before returning slowly to its steady state. The peak in consumption takes longer, and is not reached even after five years. This reflects habits in the household's utility function – which restricts the initial increase in consumption – as well as the household's desire to smooth consumption across time in response to a temporary change in income.

The increase in resource prices causes a 1.5 per cent appreciation of the real exchange rate. The real exchange rate remains elevated for a prolonged period; after five years it remains 1 per cent above its steady-state value. Despite the appreciation, export volumes increase. This is entirely due to an expansion in resource exports, although they return to their steady-state level after six years. The expansion in domestic demand and real exchange rate appreciation lead to a persistent increase in import volumes.

In the model, a temporary increase in resource prices has almost no effect on CPI inflation, although the error bands around this response are wide. The aggregate responses conceal sizeable changes in relative prices, however. To expand production, resource firms demand more labour, which increases wages and costs throughout the economy. In conjunction with the expansion in domestic demand, this leads to higher inflation in the non-tradeable and non-resource tradeable sectors. But the increase in domestic inflation is offset by the exchange rate appreciation, which lowers the inflation rate of imported items. Jääskelä and Smith (2013) and Downes, Hanslow and Tulip (2014) find that an increase in resource prices may have a small (possibly negative, at least in the short run) effect on Australian CPI inflation.

A rise in resource prices has an uneven impact on the various sectors of the economy. As one might expect, the resource sector experiences a prolonged expansion. Activity in the non-tradeable sector also increases, reflecting the pattern of domestic demand in the economy. In contrast, after a brief increase, the non-resource tradeable sector contracts. This is due largely to the appreciation of the exchange rate and an increase in firm costs, which raise the price of non-resource tradeable goods relative to goods produced overseas and lower demand for these goods in overseas markets. Consistent with the patterns of production, the increase in investment is concentrated in the resource sector; investment in the non-tradeable sector picks up three years after the increase in resource prices, while investment in the non-resource tradeable sector experiences a prolonged slump.

4.4 Variance Decompositions

In this section we examine which shocks the model suggests are the most important for the evolution of Australian economic variables.

Table 5 decomposes the unconditional variance of the observable variables into the contribution of the various structural shocks in the model. For clarity, we group the shocks into six categories. The first contains productivity shocks: the unit root (ε_µ), investment (ε_ϒ) and sector-specific (ε_n, ε_m, ε_z) shocks. The second contains demand shocks: the consumption preference shock (ε_c) and the expenditure shock (ε_g). The third contains supply shocks: the mark-up shocks in the non-traded (ε_πn), non-resource traded (ε_πm, ) and import (ε_πf) sectors. The fourth contains shocks to resource prices (ε_pz). The fifth contains domestic monetary policy shocks (ε_r). The sixth contains shocks originating abroad: the risk premium shock (ε_ψ) and the shocks to foreign output (ε_y*), inflation (ε_π*) and interest rates (ε_r*).

The model's productivity shocks explain a large proportion of the variation in the growth rates of Australian domestic demand and output, although demand shocks also explain around a quarter of output growth volatility. This is largely explained by the investment-specific technology shock, which accounts for a large proportion of the variance in investment. The contribution of the other productivity shocks is much smaller. World shocks explain around 7 per cent of the variance of export growth, but relatively little of the other demand-side variables.

Mark-up shocks are found to explain a large proportion of the variance of CPI inflation, although productivity and world shocks together explain around 20 per cent of the variance of this variable. On the production side, domestic demand shocks are estimated to explain around a third of the variance of output growth in the non-tradeable sector, but relatively little of the variation in tradeable output. The model's productivity shocks explain much of the variation in all three production sectors. Variation in the nominal exchange rate is estimated to largely be driven by world shocks – in particular, risk premium shocks.

The model suggests that resource price shocks explain relatively little of the variance of Australian macroeconomic variables, although they do explain 6 per cent of the variance of export growth, 7 per cent of the variance of the cash rate and 5 per cent of the variance of the nominal exchange rate.

World shocks are estimated to make only a small contribution to the variance of Australian macroeconomic variables other than the exchange rate and, to a lesser extent, interest rates and inflation. The modest contribution of foreign disturbances is a common finding in the open economy DSGE literature (Justiniano and Preston 2010a). In contrast, VAR models of the Australian economy typically attribute a larger share of macroeconomic volatility to foreign disturbances (for instance, see Lawson and Rees (2008) and Dungey and Pagan (2009)).

Footnotes

For comparison, the median cash rate response to a one standard deviation shock is 40 basis points on impact. [10]

In Dungey and Pagan (2009), the contraction in GNE is larger than the contraction in GDP, implying (as in our model) that net exports make a positive contribution to GDP growth. [11]

A one standard deviation shock causes a 3.4 per cent appreciation of the nominal exchange rate. [12]

Although business services industries typically have little direct trade exposure, they have a relatively high degree of indirect exposure through sales to firms in the tradeable parts of the economy. [13]

A one standard deviation shock increases foreign-currency resource prices by 5.7 per cent. [14]