RDP 9005: Real Exchange Rates and Australian Export Competitiveness 3. Technical Issues[6]

The real exchange rate indices calculated in this paper are geometric weighted averages of relative prices, adjusted for exchange rate movements.[7] The nominal exchange rate measures are calculated as geometric averages of nominal exchange rates.[8]

The base period chosen for the bilateral index is the average of the period 1980–1989. By using the average of a long period as the base, the index will not be as sensitive to short-run fluctuations, but should reflect broad trends.[9] The global export index is based in calendar year 1987. Although the selection of one individual year is not desirable, the large amounts of disaggregated data required to construct this index explains the choice of a single base year.[10]

Weights

Several different weighting schemes can be employed, based on fixed or moving weights. Moving weights ensures that the index reflects the actual pattern of trade, but changing the weights every year obscures the precise meaning of the index. On the other hand, using fixed weights may cause the index to become less relevant as trading patterns change. A compromise between the two weighting procedures can be achieved by changing the weights at fixed intervals, then splicing the new series with the old one, or by employing a system of smoothed moving weights.[11]

Five sets of weights are used in this paper, the first four being based on aggregate bilateral export data. The first set of weights used is based on raw annual export shares, whilst the second set of weights is based on a centred five yearly moving average of export shares. The third series is based on a fixed set of export weights which changes every five years, where the new series is spliced on to the old one. The fourth series is estimated using fixed weights which are an average of the annual export shares from 1980 to 1989. The final set of weights is the third country export weights, calculated by measuring the importance of each market to Australian exports, weighted by the degree of competition with other countries in each export market.[12]

The weights used for six of Australia's major trading partners under the moving average (MA) and third country weighting schemes are presented in Table 1. The weights used in the International Monetary Fund (IMF) and Australian Bureau of Agricultural and Resource Economics (ABARE) trade weighted real exchange rate indices are included for comparison.[13] This table shows that the third country weighting scheme gives substantially higher weight to Canada and the USA. These countries are significant competitors in Australia's export markets, and hence their weight in the index is greater than the export share assigned to them in the other indices. The weights given to the UK and Japan are much lower in the third country weighting scheme, because bilateral export shares overstate the importance of these countries to the competitive position of Australian exports. The export weighted indices also give a higher weight to Japan than the ABARE trade weighted index. This is because Japan is more important to Australia as an export destination than as an import supplier. The MERM index gives a higher weight to the US, because of its importance to trade between industrialized nations, and a much lower weight to Japan, since the weights are based on trade flows in 1977.

Table 1
Export and Trade Shares of Major Trading Partners in 1989
(percentage weight assigned to each country, normalized)
Country MA 3rd Country MERM ABARE
Canada 3.2 29.4 13.6 3.4
Japan 53.0 11.0 15.2 40.6
New Zealand 10.3 8.7 0.0 8.4
UK 7.4 4.4 2.9 10.3
USA 21.1 36.6 58.1 28.7
West Germany 5.0 9.9 10.2 8.6

Footnotes

Details of the construction of the indices and results are presented in the Appendix. [6]

Maciejewski (1983) shows that a nominal exchange rate index adjusted for inflation differentials is equivalent to an index of real exchange rates. [7]

In an arithmetic index, proportionate changes become dependent upon the base period selected. An arithmetic index will have an upward bias because currencies that appreciate against the Australian dollar more than the average are given a greater weight. See Brodsky (1982) for a discussion of the shortcomings of arithmetic exchange rate indices. [8]

Koch (1984) argues that the base period should reflect recent economic developments and should be cyclically neutral. [9]

Using fixed weights in a geometric averaging formula ensures that proportionate changes in the index are not dependent upon the base period chosen. [10]

In this paper, five yearly centred moving averages are used to smooth trade share weights. Figures for 1988/89 and 1989/90 are 3 and 2 year moving averages. [11]

Details of the products, markets and competitors used can be found in the Appendix. [12]

ABARE weights are 1988/89 trade weights. The IMF weights are based on MERM weights- see Artus and McGuirk (1981). [13]