RDP 9304: Exchange Rate Pass-Through: The Different Responses of Importers and Exporters 4. The Data

Estimation of a pass-through relationship requires the domestic price of a traded good, its world price and an exchange rate. In principle, the distinction between these variables is simple. However, in practice, there are a number of measurement issues that require consideration. The data used in estimation of the pass-through relationship are listed below, with a comment on their features. All data are quarterly and are expressed as logarithms. Further details and sources are given in Appendix 1.

IPDm implicit price deflator for endogenous imports, seasonally adjusted, excluding computers.
IPDx implicit price deflator for manufactured exports, seasonally adjusted, excluding computers.
WPm an index of world export prices, with weights proportional to import shares from the largest suppliers of non-oil imports to Australia.
WPx an index of world export prices, with weights proportional to world trade shares of countries that are major exporters of manufactures.
TWIm exchange rate measured by a TWI (expressed in units of domestic currency per unit of foreign currency), with weights proportional to import shares from the countries used in WPm.
TWIx TWI with weights proportional to manufactured export shares to the countries used in WPx.

Using implicit price deflators to represent the domestic price of imports and exports facilitates the requisite time series for econometric analysis. However, use of these data has several implications. First, in recent years, these deflators have been affected by rapidly falling computer prices. Thus, for the purpose of this paper, IPDm and IPDx have been adjusted to exclude computers. This is principally because the method employed by the Australian Bureau of Statistics to estimate computer prices differs to that adopted by other agencies and, therefore, detracts from the comparability of domestic and world prices.[14]

Second, implicit price deflators record prices ‘free on board’ – that is, over the docks. Furthermore, where goods are invoiced in foreign currency, their value is converted to domestic currency at the exchange rate that prevails at the time of departure rather than receipt. Thus, for imports, a shipping lag is built into the price series. No adjustment for a shipping lag has been pursued here.[15]

Third, and perhaps most importantly, a free on board price is different to a final retail price of a traded good. In fact, use of a free on board price confines econometric analysis to the first stage of pass-through – that is, the response of price setting agents in the supplying country to exchange rate changes. However, identification of first stage pass-through is a pre-requisite for examination of price adjustment at the second stage in which local distributors may or may not pass on the full cost of the traded good.[16] Moreover, quantitative analysis of the second stage of price adjustment is beyond the scope of the present paper. In consequence, second stage pass-through will be dealt with in a qualitative way in this paper.

A world price index was constructed to represent the foreign currency price of manufactured goods imported by Australia. Given that endogenous imports are principally of manufactures, an average of the foreign currency price of exports from Australia's trading partners was thought to be a suitable proxy for the world price of these goods. WPm is based on the world price index constructed by Richards and Stevens (1987).

Ideally, a world price index of the manufactures exported by Australia would comprise an average of the foreign currency price of manufactures imported by Australia's trading partners. However, appropriate data on foreign manufactured import prices are difficult to obtain. Consequently, WPx is constructed to best represent that price faced by Australian exporters of manufactures when competing in world markets.[17]

Finally, a nominal effective exchange rate index was defined. The weight given to each country in an effective exchange rate index is most often based on its total trade with the domestic economy. However, for investigating import price pass-through, such weights should be proportional to import shares (TWIm). Likewise, for investigating pass-through to manufactured export prices, weights should be proportional to shares of manufactured exports (TWIx).[18] The different weights result in TWIx depreciating by significantly less than TWIm during the mid 1980s and in recent quarters.[19]

The data are graphed in Figures 3 and 4. Note the relative stability of world prices over the period. Further, IPDm and TWIm tend to move in line with each other, suggesting that pass-through may be complete in the long run. A direct relationship between IPDx and TWIx is, however, less apparent.

Figure 3: Data for Import Price Pass-Through
Figure 4: Data for Export Price Pass-Through


Problems associated with the measurement of computer prices are discussed in Appendix 2. [14]

Phillips (1988) constructed import price series adjusted for the shipping lag, but found that it did not greatly influence estimates of exchange rate pass-through in the long run. It serves only to increase pass-through in the initial quarter. [15]

In fact, even where attempts have been made to examine second stage pass-through using industry specific data, it has been assumed that pass-through was complete and instantaneous at the first stage. (See for example Andrew and Dollery (1990).) [16]

That is, with a simplifying assumption: that those countries to which Australia exports manufactures also have access to a world market represented by the major sources of world manufactured exports. [17]

Alternatively, Menon (1992b) advocates a contract-currency weighted exchange rate. [18]

Whilst New Zealand is not a major source of Australia's imports, it is the nation's single largest destination for manufactured exports. In consequence, a substantial weight is attached to New Zealand in TWIx. Given that the Australian dollar and the New Zealand dollar tend to move together, TWIx has depreciated by significantly less than TWIm. [19]