RDP 9407: Explaining Import Price Inflation: A Recent History of Second Stage Pass-through 4. The Data

There have been a number of practical investigations of first stage pass-through in Australia.^[14] However, there has been little investigation of second stage pass-through in Australia, and indeed elsewhere.^[15] Data constraints, in particular with respect to costs and margins, have been the key factor inhibiting empirical research. In this paper, an attempt has been made to overcome some of these constraints. The data to be used in estimation are listed below, with a comment on the way in which they have been derived. For further details and sources see Appendix 1.

Central to our investigation is the retail price of imports, R. The final retail price of imports is typically represented by “items wholly or predominantly imported” contained in the CPI. However, this series is only available from the December quarter 1984, imposing constraints on econometric estimation of the pass-through relationship. A proxy for this series was constructed and spliced with the actual series. These additional observations permit estimation of the dynamics of pass-through to retail import prices during the depreciation of the mid 1980s as well as more recent episodes of major exchange rate movement.

Since final retail import prices are of consumption imports, all other variables to be used in estimation relate to such goods. A free-on-board price of consumption imports, P, was used. A published series is available for the period since the September quarter 1982. For observations prior to this date a series was constructed by removing commodity groups from the Australian Bureau of Statistic's import price index that are not predominantly of consumption goods. A landed price for consumption imports, P_L, was also constructed by accounting for the average rate of duty paid.

Similarly, an index was constructed to represent the world price of Australia's consumption imports, W. A weighted average of the foreign currency price of consumption exports from Australia's major trading partners was used. W is based on the world price index constructed by Richards and Stevens (1987), but is adapted to include the prices and volumes of consumption goods.

A nominal effective exchange rate was defined, E. Typically, the weight given to a country in an effective exchange rate is proportional to its total trade with the domestic economy. However, for investigating pass-through of exchange rate changes to the price of consumption imports, weights should be proportional to import shares of such goods.^[16]

Finally, an index of domestic costs was constructed, C. Costs other than ad valorem taxes on international trade were compiled into a composite index of domestic costs. Domestic costs for which data were available include unit labour costs, international freight, domestic transport costs and “other” expenses; they were assumed to comprise total domestic costs. An attempt was made to apportion these domestic costs according to their share in retailers' total costs.^[17] Certainly, C does not represent all of the domestic costs incurred in bringing an import to a retail outlet. However, those items excluded (such as property rental and charges for utilities), fluctuate less frequently. In consequence, major changes in costs faced by importers are likely to be captured by movements in C.

In the following sections, these data will be employed to investigate first and second stage pass-through.

Footnotes

See Dwyer et al. (1993), Menon (1991b, 1991c, 1992a and 1992b), Lattimore (1989), Phillips (1988), Coppel, Simes and Horn (1987) and Richards and Stevens (1987). [14]

One of the first attempts was by Andrew and Dollery (1990). However, they do not actually isolate the second stage of price adjustment. Instead, they look at the combined pass-through of a change in the exchange rate to the retail price of Australian passenger motor vehicles. [15]

Whilst, in principle, the choice of trade weight matters, for practical purposes there is little difference between an effective exchange rate based on shares of consumption imports compared with aggregate imports (see Appendix 1). In other exercises, however, the choice of trade weight does matter. See, for example, Dwyer et al. (1993, pp. 10–11). [16]

Various census and survey information about the retail industry permits identification of the contribution of key costs items to the final retail price of an import. With a few simplifying assumptions, the corresponding shares of retailers' total costs can be inferred. Details of this apportionment are described in Appendix 1. [17]