RDP 2014-11: Exchange Rate Movements and the Australian Economy 3. Results

3.1 Aggregate Model

We first discuss aggregate responses to an unanticipated exchange rate appreciation, shown in Figure 2, before moving to the sectoral implications. In all examples, we scale the initial shock so that it causes a 10 per cent appreciation of the real exchange rate.[8] After the initial shock, the real exchange rate evolves endogenously, although we also show the effects of a more sustained exchange rate appreciation.

Figure 2: Impulse Response Functions – Exchange Rate Shock

The appreciation causes a persistent contraction in GDP.[9] The point estimates suggest that the maximum effect of the appreciation occurs around eighteen months after the shock, at which time the level of GDP is 0.3 per cent below its trend level. Although the confidence intervals around this response are wide, the decrease in GDP is statistically significant at a one standard deviation confidence level after three quarters. Subsequently, output returns to trend, which it reaches four years after the initial appreciation. The appreciation also triggers a fall in domestic inflation. In year-ended terms, the inflation rate is around 0.3 percentage points lower eighteen months after the initial appreciation before gradually returning to trend in subsequent quarters.[10] The cash rate responds little to the shock in the first few quarters. This accords with existing empirical evidence that exchange rate movements have little direct effect on Australian monetary policy. Monetary policy responds to the decreases in output and inflation, however, and the cash rate is slightly more than 40 basis points lower than otherwise, before it returns to its initial level.

In our baseline results we allow the real exchange rate to evolve endogenously following the shock. In this case, the exchange rate appreciation is relatively short-lived, with the real exchange rate returning to its initial level within two years of the shock. To illustrate the implications of a more persistent appreciation, Figure 3 shows the response of GDP when we impose a sequence of exchange rate shocks so that the real exchange rate remains 10 per cent above its initial level for three years before returning endogenously to its initial level. In this case, the contraction in output is more prolonged; output troughs at around 0.8 per cent below its initial level, three years after the initial appreciation, before recovering gradually.

Figure 3: GDP Response to Temporary and Persistent Exchange Rate Shocks

3.2 Production Sectors

Figure 4 shows how the output of different industries responds to a temporary 10 per cent exchange rate appreciation. Consistent with the decrease in aggregate output following an exchange rate appreciation, most industries experience a reduction in activity. The mining and manufacturing industries experience much larger decreases in output than the economy as a whole. The response of mining occurs in the quarters immediately following the shock, while it takes three to four quarters before the manufacturing sector contracts substantially. Output in the personal services and other business services sectors also decrease substantially following the shock. In contrast, the model suggests that output in the construction and goods distribution sectors increases initially following an exchange rate appreciation, although these sectors contract later on.

Figure 4: Industry Gross Value Added

The appreciation appears to have a large positive impact on output in the finance & insurance services industry. We find it difficult to explain this last result and suspect that it may reflect measurement issues, such as the structural changes that occurred in this industry during the deregulation of Australian financial markets during the 1980s. Overall, the results illustrate how the aggregate effect of the exchange rate shock masks significant differences in responses across industries.

A number of factors might influence the magnitude of an industry's response to exchange rate shocks. Most directly, the trade exposure of the industry, either in terms of its reliance on foreign demand or the extent of import competition, is likely to matter. In industries that are more trade-exposed, an exchange rate appreciation is likely to cause a larger and more rapid decline in demand as expenditure switches towards foreign output. Working in the opposite direction is the effect of exchange rates on production costs. If an industry relies heavily on imported intermediate inputs then an exchange rate appreciation will lower costs. This may allow firms in that industry to lower their prices and increase sales. Finally, the degree of competition in the industry, which is likely to be related to average firm mark-ups, may also determine whether firms adjust margins or prices in response to exchange rate movements. In industries that adjust margins, we would expect exchange rate movements to have a smaller absolute effect on output.

To disentangle these different channels, Table 1 lists descriptive statistics for each industry in the Australian economy.

Table 1: Industry Characteristics
Per cent
  Share of output(a) Average mark-up(b) Export propensity(c) Imported input share(d) Import penetration(e) Trade-related exposure(f)
Mining 4.3 56½ 4 11½ 30½
Manufacturing 7 1.4 20 15 34½ 29¼
Construction 1.8 ¼ 4 0
Goods distribution 14½ 1.3 11½ 20
Wholesale trade 1.2 0 22¼
Retail trade 5 1.2 3 0
Transport, postal & warehousing 2.4 19¼ 8 26¼
Finance & insurance services na 1 ½ ¾ 9
Other business services 16 2.3 3 3 12½
Information media & telecommunications 3 3.0 3 6
Rental, hiring & real estate services 3.5 ¾ ¾
Professional, scientific & technical services 1.9 3 16¼
Administrative & support services 3 1.6 2 4 25½
Personal services 1.8 10¾
Accommodation & food services 1.7 11¾ 6
Arts & recreation 1 3.5 4 10¼
Other services 2 1.6 ½ 12¼ ¾ 22
Social services 11¾ 1.7 ½
Education & training 5 1.5 1
Health care & social assistance 7 1.8 ½ 5 ½ 0
Agriculture, forestry & fishing 2.2 14 2 61
Electricity, gas, water & waste services 3 2.0 0 0 18¾
Public administration & safety 2.0 ¼ 0

Notes: (a) Industry gross value added (GVA) as a share of aggregate GVA; data for 2012/13
(b) Sum of sales and service income plus funding from government plus change in inventories divided by the sum of labour costs plus purchases; data an average of 2001/02 to 2011/12
(c) Exports as a share of total domestic production in each industry; data for 2009/10
(d) Imported inputs as a share of total domestic production in each industry; data for 2009/10
(e) Competing imports as a share of the total supply of the corresponding industry; data for 2009/10
(f) Percentage of domestic sales to an industry with an export propensity of greater than 10 per cent; data for 2009/10

Sources: ABS; Authors' calculations

As expected, highly trade-exposed industries like manufacturing and mining are among the most responsive to exchange rate movements. However, large responses are not limited to the parts of the economy with direct trade exposure. For example, business services industries generally cater to the domestic market, have little imported competition and yet display a large response to exchange rate movements. Many of these industries do, however, have considerable exposure to trade-exposed industries.[11] For example, around a quarter of the output of administrative & support services is ultimately consumed by trade-exposed industries. So, while business services firms typically have little direct trade exposure, they have considerable indirect exposure to foreign trade. This may explain their responsiveness to exchange rate movements.

Evidence of a cost channel of exchange rate transmission is limited. Many sectors with a relatively large imported input share, including the manufacturing and personal services industries, experience relatively large decreases in output following an exchange rate appreciation that reduces their cost of imported inputs. Given the trade exposure of these two industries, it is likely that the expenditure-switching effect overwhelms any benefits that these industries receive from lower costs. In contrast, lower imported input costs may explain the initially large positive response of goods distribution output.[12]

The relationship between industry mark-ups and responsiveness to exchange rate movements is difficult to discern in the data. For example, although average mark-ups are relatively low in the manufacturing and goods distribution industries, the responses of these two sectors to exchange rate movements differs greatly. It may be that mark-ups affect these two industries differently. Low mark-ups in the manufacturing industry may prevent firms from reducing the Australian dollar prices of their output in response to an exchange rate appreciation. This may contribute to the large decreases in output that this industry experiences following an appreciation. In contrast, if there is a relatively high degree of competition in the goods distribution sector, then firms may be forced to pass on to customers a large portion of any reduction in the costs of goods sold. This may contribute to the initial increase in output that this industry experiences. However, it is not possible to quantify these effects separately using our model.

One might wonder whether our sectoral responses, which are calculated using a separate VAR for each industry of the economy, are consistent with responses of the aggregate model. To answer this question, Figure 5 compares the response of GDP in our aggregate model with the cumulated responses of the sectoral models, with each industry's output weighted by its average share of output. Reassuringly, the two lines correspond extremely closely and in both the appreciation triggers a 0.3 per cent decrease in the level of GDP after 12–18 months. The response of the sectoral models appears to lag the responses of the aggregate model by a quarter or so. However, given the fairly large confidence bands around the aggregate responses, this difference is not statistically significant.

Figure 5: Gross Domestic Product

3.3 Is the Exchange Rate a Shock Absorber or a Source of Shocks?

The previous section showed that real exchange rate shocks can have a meaningful impact on the Australian economy. This raises the question of whether real exchange rate shocks (that is, those exchange rate movements that are not an endogenous response to changes in other economic variables) are an important cause of Australian macroeconomic volatility.

It turns out that the answer to this question is no. To demonstrate this, Table 2 shows the contribution of real exchange rate shocks to the variance of other domestic variables in the VAR at horizons ranging from six months to ten years.[13] Regardless of the horizon, exchange rate shocks explain little of the volatility in any of the other domestic aggregate variables.

Table 2: Variance Decomposition – Aggregate Variables
Per cent of variance explained by exchange rate shocks
Variable Horizon (quarters)
  2 10 20 40
GDP 0.2 2.6 2.1 1.5
Inflation 1.0 3.6 3.9 3.3
Cash rate 0.1 1.9 3.8 3.2

At an industry level, the pattern is similar (Table 3). Exchange rate shocks explain a modest proportion of the variance of the manufacturing and other business services sectors at horizons of more than two years. Surprisingly, the VARs also suggest that exchange rate shocks contribute to the variance of the electricity, gas, water & waste industry. Although this industry has little direct exposure to foreign trade, it does have a relatively high indirect exposure through its interactions with trade-exposed industries (Table 1). However, exchange rate shocks explain little of the variance of other industries in the economy.

Table 3: Variance Decomposition – Industry Variables
Per cent of variance explained by exchange rate shocks
Variable Horizon (quarters)
  2 10 20 40
Agriculture, forestry & fishing 0.9 1.6 2.0 2.0
Mining 0.6 1.2 1.1 1.0
Manufacturing 0.4 7.2 5.9 4.4
Electricity, gas, water & waste services 0.1 5.2 5.2 4.9
Construction 0.1 0.5 0.5 0.5
Finance & insurance services 0.2 3.6 2.2 1.5
Other business services 0.1 5.8 6.5 5.4
Personal services 0.0 3.4 2.7 1.6
Goods distribution 5.4 4.3 3.2 1.8
Social services 0.0 0.2 0.4 0.3

Variance decompositions can also be used to examine what shocks typically cause real exchange rate movements. Table 4 shows the contribution of foreign and domestic shocks to the variance of the real exchange rate. At short horizons, much of the variance of the real exchange rate is determined by its own shock. This is consistent with the observation that movements in nominal exchange rates (which determine most short-horizon movements in the real exchange rate) cannot be explained by macroeconomic fundamentals (Meese and Rogoff 1983). At horizons beyond two years, foreign variables are the most important determinant of real exchange rate fluctuations. This aligns with previous research that has shown that long-run movements in Australia's real exchange rate can largely be explained by the evolution of the terms of trade as well as interest differentials between Australia and overseas (Gruen and Wilkinson 1994), though it contrasts with more recent findings by Voss and Willard (2009) that attribute only a small share of Australia's exchange rate movements to foreign variable shocks. Other domestic shocks explain only a small proportion of the movements in the real exchange rate, and their influence is greatest over a horizon of around two years.

Table 4: Variance Decomposition – Exchange Rate
Per cent of variance
Shock Horizon (quarters)
  1 10 20 40
Foreign 11.3 60.1 82.0 92.0
Domestic non-exchange rate 2.6 11.8 5.4 2.4
Exchange rate 86.1 28.1 12.6 5.6

In sum, our model indicates that over all but the shortest horizons Australia's real exchange rate is largely determined by international factors and that exchange rate shocks contribute little to the volatility of other macroeconomic variables in Australia. These results provide some support to the idea that exchange rate movements have served to mitigate the impact of foreign shocks on the Australian economy, rather than acting as a source of shocks in their own right.[14] We investigate this issue further in the following section.

Footnotes

Since our model is linear, the impacts of an appreciation and a depreciation are symmetric. In reality, large exchange rate movements may have greater-than-proportional impacts due to their effects on firm entry and exit. In our aggregate model, a one standard deviation exchange rate shock causes a real exchange rate appreciation of around 3 per cent. [8]

We generate confidence intervals using Killian's (1998) bias-adjusted bootstrap method. [9]

Although we include quarterly inflation in the VAR, we present the results in year-ended terms for ease of interpretation. [10]

For a more detailed discussion of supply chains in Australia, see Kelly and La Cava (2014). [11]

Although the goods distribution sector's imported input share is only around 5 per cent, this figure only accounts for the value of distribution services. That is, it does not include the actual value of the goods sold by wholesalers and retailers. However, demand for distribution services is linked to the final prices of the goods sold, which in turn are linked to the cost of purchasing these goods. We estimate that the imported input share of the goods distribution sector rises to around 19 per cent when the value of the goods sold are included as both an input and an output of the sector. [12]

The assumption of block exogeneity implies that exchange rate shocks do not contribute to the variance of the foreign variables in the VAR. [13]

This notion typically emphasises flexibility of the nominal, rather than real, exchange rate. However, price stickiness means that, over short horizons, nominal exchange rate flexibility determines the degree of real exchange rate flexibility. [14]