RDP 2016-05: Trade Invoicing Currency and First-stage Exchange Rate Pass-through 2. Import Invoicing Currencies in Australia

We construct time series of import invoicing currencies at the two-digit Standard International Trade Classification (SITC) level. We do so using data from the ABS International Trade in Goods and Services (ITGS) data release and, prior to March 2003, the ABS International Merchandise Trade (IMT) data release.^[3] Data on invoice currencies are unavailable for services imports so we consider only goods imports, as do Gopinath et al (2010).

The ABS publishes data on invoice currencies for exports and imports at the two-digit SITC level roughly every two years. Table 1 shows the underlying surveys from which we draw the data, and the quarters that they cover.

The ABS provides data on the average invoice share across the survey period for a number of currencies at the two-digit SITC level. We construct quarterly two-digit SITC-level invoice share series as follows:

• For quarters during the survey we take the survey period average.
• For quarters falling between surveys we linearly interpolate between the last survey reading and the next.

The ABS also publishes aggregate invoice share data for each quarter within the survey period. We do not use these aggregate-level data in our regression analysis (but they are shown in Figure 1).

In total, we have invoice currency data for six currencies across 23 two-digit SITC divisions. Based on the most recent weighting for the import price index, these 23 divisions cover a little under 85 per cent of all goods imports. The six currencies cover around 98 per cent of imports in the SITC divisions for which we have data.^[4]

2.1 Trends in Invoice Currencies

The majority of Australia's goods imports are invoiced in foreign currency, mostly US dollars (Figure 1). The average proportion of imports invoiced in US dollars has held reasonably steady at a little over 50 per cent across most of the period for which we have data. The share has increased a little in recent years, matching a similar-sized decline in the share of imports invoiced in Australian dollars. As of the most recent period for which we have data on currency of invoice, 57 per cent of all goods were invoiced in US dollars, 29 per cent in Australian dollars, 8.3 per cent in euro, 1.3 per cent in yen, 1.1 per cent in British pounds and 1 per cent in New Zealand dollars. The ABS does not routinely provide invoice currency data for other currencies.

The share of goods imports invoiced in US dollars is noteworthy given that only 11.5 per cent of Australia's goods imports come from the United States. This may in part reflect that: many of Australia's Asian trading partners have exchange rates closely linked to the US dollar; the US dollar is used as a global reserve currency; and the US dollar is used as a reference currency for many international prices – for instance, many commodities are routinely quoted in US dollars.^[5] Goldberg and Tille (2008) argue that exporters often choose to invoice in a common currency to minimise variation in prices relative to their competitors, which they refer to as a ‘coalescing effect’. Critically, exporters may coalesce around an invoicing currency other that used in the importing or exporting country, as is the case for much of Australia's US dollar-denominated trade.

2.2 Invoice Currency by Broad Product Category

Table 2 shows the trade invoicing currency shares at the two-digit SITC level, as at June 2014 (the last period for which we have invoice currency data). The share of Australian dollar-invoiced trade is generally high for differentiated goods and low for homogenous goods. For example, petroleum products (SITC 33) are invoiced almost entirely in US dollars. Similarly, relatively homogenous manufactures – such as iron and steel (SITC 67) and non-ferrous metals (SITC 68) – are largely invoiced in US dollars. Differentiated goods, such as pharmaceuticals (SITC 54) and vehicles (SITC 78), tend to be invoiced in Australian dollars. Clothing (SITC 84) is an exception to this pattern, for which the share of imports invoiced in US dollars is high despite being a division that covers differentiated goods.

2.3 SITC-level Exchange Rate Indices

For use in the regression analysis that follows, we construct 19 quarterly two-digit SITC-level import trade-weighted exchange rate indices (Figure 2).^[6] These indices are designed to measure the most relevant exchange rate changes for each two-digit SITC division because the composition of the countries of origin differs across goods imports at the two-digit SITC level. Thus, some bilateral exchange rates are more relevant for some goods than others. For instance around a third of road vehicles imports (SITC 78) come from Japan, compared to about 1 per cent of medicinal and pharmaceutical imports (SITC 54); thus, the Australian dollar-Japanese yen exchange rate is more relevant for the former than the latter.

These indices are geometric weighted averages of bilateral nominal exchange rates, with the weights determined by the import trade share of the relevant country at the two-digit SITC level. The weights are the 12-month moving average of the country's share of total imports within each two-digit SITC division; we take a moving average to smooth month-to-month variation in trade shares, which can be affected by lumpy imports. That is:

where:

• is the weight of country c for SITC division i at month t
• is imports from country c for SITC division i at month t − j.

We use the monthly ABS balance of payments two-digit SITC-level trade data for imports by country to construct these weights. Although we construct these indices at monthly frequency, quarterly averages are taken to match the frequency of other data used. All the indices are indexed to September 1996. Appendix A examines some alternative specifications for the two-digit SITC-level import trade-weighted exchange rate indices; our results are robust to these alternative ways of measuring the relevant exchange rate.

There is a tight relationship between the import trade-weighted exchange rate indices we construct and the aggregate import trade-weighted exchange rate index. Nonetheless, there are differences between the series, particularly in quarterly changes. This variation in exchange rate changes across SITC divisions is helpful for identifying exchange rate pass-through. Nonetheless, most of our identification of pass-through comes from the sizeable time series variation in exchange rates common to all SITC divisions.^[7]

2.4 Import Invoicing and Pass-through

Mechanically, contemporaneous first-stage pass-through is complete for imports invoiced in foreign currencies because the Department of Immigration and Border Service converts foreign currency-invoiced imports to Australian dollars using exchange rates that apply at the time of import. Accordingly, there is a strong contemporaneous relationship between import price indices for SITC divisions with high proportions of foreign currency invoicing and the SITC-level import trade-weighted exchange rate indices that we construct. For example, apparel and clothing imports are largely invoiced in foreign currencies (Figure 3), and the relationship between movements in the SITC-level import trade-weighted exchange rate index and the import price index for these goods is close to one-for-one (Figure 4).

For Australian dollar-invoiced goods imports the contemporaneous relationship is much weaker. For instance, the import price index for road vehicles only loosely follows the SITC-level import trade-weighted exchange rate index (Figure 5). Moreover, the relationship looks like it may have weakened over time. This weakening might reflect the growing share of Australian dollar-invoiced goods in this category: 75 per cent of all imported road vehicles are currently invoiced in Australian dollars, up from 50 per cent in the late 1990s (Figure 6).

Figures 3 and 6 also show that there is substantial variation in the prevalence of Australian dollar invoicing across two-digit SITC divisions. It is this variation that we exploit to estimate how invoicing currency affects first-stage pass-through.

Footnotes

The Asian economic crisis stimulated interest in the currencies used to invoice Australia's trade, prompting the ABS to begin collecting trade invoicing currency data in 1997. [3]

A full list of these SITC divisions is provided in Table 2. [4]

For comparison, around 90 per cent of Canadian imports are invoiced in US dollars (Devereux, Dong and Tomlin 2015); in the United Kingdom around two-thirds of non-EU imports are invoiced in US dollars, and around a quarter are invoiced in British pounds (HM Revenue & Customs 2015). [5]

We exclude four SITC divisions comprising mostly homogeneous goods (e.g. petroleum products) in the regression analysis that follows. See Table 2 for a list of the SITC divisions we have data for, and those that we exclude from the regression analysis. [6]

Around four-fifths of the variation in the two-digit SITC-level exchange rate indices is common. [7]