RDP 9106: The Direction of Australian Investment from 1985/86 to 1988/89 6. Tradeable and Non-Tradeable Investment in the Manufacturing Sector

The objective in this section is to obtain a better measure of the level of investment in the tradeable sector over the period of the most recent investment boom. Interest in investment in the tradeable sector derives from the view that meeting the challenge of Australia's external imbalance requires structural change, i.e. it is necessary for manufacturing industries to make new inroads into international export markets and to compete more effectively against imported manufactures. This view implies that the “right” industries for investment to be going towards are those industries which are tradeable.

In principle a tradeable industry is one where a movement in international relative prices favouring its product would see substitution in demand and supply towards producing for export (or import competition). It would be interesting to know whether the sharp fall in Australia's real exchange rate from 1985 was associated with investment in those industries where these price elasticities are greatest. Estimation of the appropriate elasticities at such a disaggregated level over the sample period chosen is not, however, practical. Most of the existing Australian literature in this area, therefore, has adopted the practice of using average (as opposed to marginal) export propensities. This may be a guide to the extent the average share of exports in production or sales of the industry is related to its price responsiveness. While this approach may be imperfect, we nevertheless present results using the methodologies of the Bureau of Industry Economics and an extension of Wood et al. (1990).

The BIE (1990) methodology involves estimating import penetration ratios and export propensities and then using a bench-mark for these ratios to categorise the industry. Where a particular industry is classified as tradeable then the total investment of that industry is deemed as tradeable. The BIE approach has two drawbacks. The first of these is that it attributes all the investment as tradeable when a significant proportion of some tradeable industry's output may be neither in the export market nor subject to import competition.^[13] The other drawback is the arbitrary nature of any bench-mark. The methodology of Wood et al. (1990) overcomes these weaknesses but they restricted themselves to an analysis of only export creating investment.

In order to develop our preferred procedure we assume that adjustment can occur on all margins, and not just the trade margin. For exportable industries, investment will lead to the production of units of output, some of which will be consumed domestically and the rest exported; the split will be captured by weighting investment by the export propensity.^[14] A similar argument also applies to industries producing importable goods. Output produced by new investment will compete with both imports and existing domestic production. We assume the extent to which output from this new investment will replace imports can best be captured by the import penetration ratio.

Like Wood et al. (1990), we weight investment in each year with the export propensity (or import penetration ratio) of that year and thus neglect dynamic problems associated with lags between when investment is undertaken and when output from this investment is produced.

There is an additional problem due to the degree of aggregation. Using a bench-mark to classify an estimation industry as tradeable or non-tradeable neglects the range of industry types that are aggregated to form this estimation industry. Thus to claim all of the investment within the estimation industry is tradeable or non-tradeable is likely to be incorrect. A priori it is not possible to determine the direction of the bias over all estimation industries, though this bias can be minimized by using data at the most disaggregated level possible (as we have done).

The aggregation problem is partially avoided by using our preferred methodology – that is, weighting investment by import penetration ratios and export propensities. The implicit assumption is that the capital to output ratio is constant across all industries within each estimation industry.

(a) Results using the BIE Approach

Export propensities and import penetration ratios for 18 of the 22 manufacturing estimation industries were calculated from unpublished ABS trade data at the ASIC 4 digit level. Results are reported in Appendix 2, Tables 6 and 7 respectively.^[15] The export propensities and import penetration ratios are those defined by the BIE (1990) as in Section 5(b). The domestic production series is from the ABS Stocks, Manufacturers' Sales and Expected Sales survey.^[16]

We classify an industry as tradeable if the appropriate ratio is greater than or equal to the bench-mark for every year of the period studied. We call these the 10-per-cent-every-year and 20-per-cent-every-year bench-marks. Table 3 and Table 4 classify the estimation industries in the manufacturing sector using these bench-marks.

A comparison of Tables 3 and 4 reveals that the traded/non-traded split is surprisingly insensitive to a change in the bench-mark. Only four industries are reclassified from being tradeable industries to non-tradeable industries as we move from the 10 to the 20 per cent bench-mark.

Using either bench-mark, we calculate the proportion of total investment which occurred in those estimation industries defined as export industries and/or import competing industries. This provides two measures of the proportion of total manufacturing investment in the traded sector. Graph 6 shows the results using both bench-marks as well as results using a preferred measure which will be defined in Section 6(b).

Not surprisingly, the less restrictive the bench-mark the higher the proportion of manufacturing investment in the tradeable sector. Results for the 10 per cent bench-mark show that the proportion of manufacturing investment in the tradeable sector remained almost unchanged over the period. For the 20 per cent bench-mark, the share of manufacturing investment in tradeable industries fell from 1984/85 to 1986/87 then rose through to 1988/89. These results show a small fall in the share of manufacturing investment directed to the tradeable sector over the period 1984/85 to 1988/89 (falls of 1.2 and 2.8 percentage points for the 10 per cent and the 20 per cent bench-marks respectively).^[17]

(b) Results using the Preferred Approach

To derive our preferred measure of the proportion of total manufacturing investment in the tradeable sector we extend the Wood et al. approach by calculating the proportion of total manufacturing investment in import competing investment. However, the definition of trade ratios used by the BIE and Wood et al. (1990) is inappropriate for this application as the import penetration ratio does not share a common denominator with the export propensity.

To determine the tradeability of investment we need to determine what will happen to output created by this investment. There are three possible “destinations”:

1. some proportion of the new output could be exported;
2. some proportion of the new output could replace imports; or
3. some proportion of the new output could be absorbed domestically
   – that is, it could replace existing domestic output or be absorbed
   by increased demand for the domestically produced output.

For each good type i (defined as being produced by estimation industry i), we have data on

The export propensity we redefine as

and the import penetration ratio we redefine as

The use of this particular denominator is made clearer by defining D_i = S_i − X_i as the domestic absorption of output that is produced domestically. Thus:

which is simply the sum of the volumes of the three possible “destinations” for new output.

The preferred measure of the proportion of total manufacturing investment in total tradeable capacity^[18] is

The results for this measure and its two components are given in Table 5 for the years 1984/85 to 1988/89; the results for this preferred measure are reproduced in Graph 6.

Our results using the preferred approach show a rise in the share of total manufacturing investment in total tradeable capacity from 33.0 per cent in 1984/85 to 37.5 per cent in 1988/89. Most of this rise can be attributed to a rising proportion of total manufacturing investment in import replacement capacity, with the strongest gain from 1987/88 to 1988/89. The proportion of manufacturing investment in export creating capacity remained fairly constant over the period; as expected this measure was slightly lower than the same calculated using the Wood et al. definition of the export propensity, though both moved together.

An examination of the relative prices of manufactured exports and imports to the GDP deflator^[20] (Graph 7) suggests an explanation for the results in Table 5. Relative manufactured export prices remain fairly flat from the beginning of 1981/82 through to the end of 1988/89, while relative manufactured import prices rose significantly from 1984/85 and stayed high until 1986/87. Thus the share of investment in import replacement capacity appears to be reacting to changes in relative prices with a lag of a few years.

Changes in the proportion of manufacturing investment in total tradeable capacity can be broken into two parts – that due to changes in trade ratios for each industry and that due to changes in relative growth rates of investment between traded and non-traded sectors.^[21] Our results imply that investment in tradeable capacity rose mainly due to a relatively higher rate of growth of investment in those estimation industries which were relatively tradeable, rather than an increase in the tradeability of industries – and this is true for both import competing and export creating investment.

Footnotes

While for a domestic industry competing with imports, all units of domestic output are thought of as competing with imports, the issue here is to what extent new investment will reduce imports. [13]

Those units consumed domestically may be as a result of growth in domestic demand and/or crowding out of existing production. [14]

Though we report import penetration ratios and export propensities for the fruit and vegetable products estimation industry (CAPEX 04), no investment data is available for this estimation industry. Thus we also include this estimation industry in the aggregated food and tobacco industry for which investment data is available. [15]

Australian Bureau of Statistics (1990a), Cat. no. 5629.0. [16]

The BIE apply bench-marks on the basis of the values of penetration ratios and propensities for only one year – 1986/87. Another alternative would be to use averages over the whole period to classify industries. We have established that results for the 10 per cent bench-mark are insensitive to changes in the way the bench-mark is applied. Applying the 20 per cent bench-mark in alternative ways (either to the average of ratios over the whole period or to the 1986/87 ratios) shifts the curve up (by about 7 percentage points) with minimal change to its shape. The 20 per cent bench-mark applied in 1986/87 shows a 1 percentage point rise in the share of manufacturing investment directed to the tradeable sector from 1984/85 to 1988/89. [17]

Note that we distinguish this measure from the previous measure of total manufacturing investment in the tradeable sector. [18]

We also calculated the proportion of total manufacturing investment in export creating investment using the Wood et al. definition of the export propensity. The results for each of the years 1984/85 to 1988/89 are: 18.2, 19.5, 17.6, 19.2, and 19.0 per cent. [19]

Manufactured export and import implicit price deflators are derived from Australian Bureau of Statistics (1990b), Cat. no. 5302.0. [20]

The algebra is similar to Section 5(c). Note that for discrete changes the breakdown will depend on the choice of base years for investment and trade ratios. In this case the general results remain unchanged. [21]