RDP 2016-05: Trade Invoicing Currency and First-stage Exchange Rate Pass-through 5. Implications

5.1 Real Rigidities

Less-than-complete pass-through for goods invoiced in Australian dollars is likely to be a result of foreign exporters with a low desired degree of pass-through choosing to invoice in Australian dollars, because this minimises pass-through for the duration over which prices are fixed (Gopinath et al 2010). Thus, currency of invoice serves as a sufficient statistic for firms' desired degree of long-run pass-through. A deeper question is why desired long-run pass-through is much less than complete for some firms. Rationalising low levels of desired pass-through amounts to a search for sources of real rigidity – factors that attenuate the response of prices to nominal shocks. Our sense of the literature (discussed in the introduction) is that, while a range of hypotheses have been explored, the low degree of long-run pass-through for local currency-invoiced goods remains a puzzle.

Gopinath and Itskhoki (2010) find some evidence that pass-through is affected by the degree of competition among importers: sectors with a small number of large importers tend to show lower levels of exchange rate pass-through, although there is considerable uncertainty around their estimates. This fact suggests that SITC divisions with a small number of large importers would be more likely to invoice in Australian dollars. Price-level evidence from Canada supports this contention, with Devereux et al (2015) finding a negative relationship between importer market share and pass-through. Consistent with Gopinath and Itskhoki (2010), Australian dollar invoicing does appear to be more common in the manufacturing industry; however, we caution that because we do not have firm- and price-level data, this invoicing-share calculation is extremely rough. Another strand of the literature notes the ongoing nature of importer-exporter relationships, with Gopinath and Itskhoki (2010) arguing that substantial mark-up variation can be generated in a bilateral bargaining environment.

In recent work, Berger and Vavra (2013) find that the degree of exchange rate pass-through is time-varying, and tends to be high when the cross-sectional dispersion of price changes across items is high, but they do not explain the low level of exchange rate pass-through. Lewis (2016) uses price-level data for imports and shows that first-stage pass-through is nonlinear at the micro level: there is little pass-through of small exchange rate changes, but pass-through of large exchange rate changes is about four times as high. Hedging practices allow firms to manage cash flow positions, but cannot explain why firms choose to set time-varying mark-ups.

Firms that operate across borders also affect the degree of first-stage pass-through because their imports are not arms-length. Neiman (2010) finds that intra-firm prices are characterised by more frequent price adjustment and higher long-run pass-through. A different but related possibility is that pass-through differences in part reflect pricing agreements for intra-firm trade reached between firms and the tax authority. The two trade divisions in our sample for which there may be considerable scope to manipulate intra-firm prices are medicinal and pharmaceutical products (SITC 54) and road vehicles (SITC 78). However, our findings are essentially unchanged if we exclude these two divisions from the estimation sample.

5.2 Monetary Policy

Local currency (Australian dollar) pricing of imports dampens the response of importers' costs to exchange rate changes, and so is likely to weaken the relationship between exchange rate changes and consumer prices. Devereux and Engel (2003) argue that this weakens the stabilisation role of floating exchange rates. Standard models featuring price stickiness and producer currency pricing imply that exchange rate changes have an immediate impact on relative prices of domestic- and foreign-produced goods, and so have large expenditure-switching effects; under certain assumptions, monetary policy under a flexible exchange rate regime with producer currency pricing can replicate the equilibrium of the economy with fully flexible prices. But with price stickiness and local currency pricing, changes in monetary policy have less influence on the relative price of domestic and foreign goods through movements in the exchange rate.

5.3 Modelling Import Prices

For the approximately 30 per cent of goods imports invoiced in Australian dollars, we have estimated pass-through of only 14 per cent after two years. This implies large and persistent deviations from the law of one price for a sizeable share of Australia's goods imports. The speed of exchange rate pass-through for Australian dollar goods imports may rise beyond the two-year horizon we have examined, but Gopinath et al (2010) do not find this for US data. In contrast, pass-through is immediate and complete for foreign currency-invoiced imports. Our results imply that the speed of adjustment in an ECM framework for exchange rate changes to import prices would be very slow for Australian dollar-invoiced trade. Furthermore, because there are two distinct sets of goods for which pass-through differs so greatly, aggregate import price models that impose the law of one price are likely to be unsuitable, at least over forecast horizons relevant for monetary policy.

5.4 Terms of Trade

Empirically, and in theoretical models, the terms of trade is a determinant of the exchange rate (e.g. Hambur et al 2015). However, our results imply that exogenous exchange rate changes might have a long-lived effect on the terms of trade. The low degree of exchange rate pass-through to Australian dollar-invoiced imports means that aggregate import prices move less than one-for-one with the exchange rate, out to at least a two-year horizon. Our estimates imply that a 1 per cent exogenous depreciation (i.e. not due to changes in the terms of trade) in the Australian dollar against all trading partners raises Australia's goods and services import price index by about 0.8 per cent after two years.^[10] If pass-through for goods exports is complete, or close to, we would thus expect a depreciation to increase the goods terms of trade.^[11] Similarly, an exogenous exchange rate appreciation would decrease the terms of trade.

Footnotes

This calculation uses our estimate of 14 per cent long-run pass-through for the 29.4 per cent of Australia's goods imports invoiced in Australian dollars, and assumes complete pass-through for other goods and services imports. [10]

Although data on goods export invoicing currencies are available, we have not pursued assessing pass-through because many of Australia's goods exports are homogenous goods, and thus the law of one price is likely to hold. [11]