RDP 9707: Internationalisation and Pricing Behaviour: Some Evidence for Australia 3. Long-term Trends in Relative Prices and Openness

In this section we examine long-term movements in the relative output prices of domestic and foreign industries and discuss a range of influences on these relative prices, one of which is internationalisation of the domestic economy.

3.1 Defining and Interpreting Movements in Relative Prices

Industry real exchange rates can be calculated by comparing the price of an industry's output with the price of output of the same industry in the rest of the world, expressed in a common currency. In conducting these calculations, account must also be taken of tariffs and transportation costs. In symbols, the relative price of a domestic industry's output and the foreign industry's output (sold in the domestic country) is given by:

where pt is the domestic price of a traded good, Inline Equation is the world price of the same traded good, et is the exchange rate, tarifft is the ad valorem tariff rate (or more generally the rate of protection accorded domestic producers from all sources of trade barriers), and transt is the per unit cost of international freight. This relative price is sometimes interpreted as a measure of competitiveness, as all else constant, a lower domestic price relative to the world price should mean that the domestic industry is more competitive in world markets.

Theory suggests that for a small open economy trading homogeneous goods, the prices of tradeable goods should be determined by the prices prevailing in the rest of the world. That is, the ‘law of one price’ should hold, after controlling for the effects of trade barriers and transportation costs. Under these assumptions, the above price ratio should always be equal to unity for tradeable goods.

In many cases, however, the assumptions underlying the law of one price do not hold. Markets may not be completely open, goods may not be perfect substitutes, and frictions in the economy may imply that it takes time for consumers and producers to adjust to shocks. In these cases, domestic suppliers or domestic factors of production may have some capacity to exploit their market power, driving the domestic price away from the level implied by foreign prices.[6]

The focus of this section of the paper is on the impact that increasing the openness of markets has on these relative price ratios. From a partial equilibrium perspective, allowing greater import competition should lead the price of domestically produced import-competing goods to either stay constant or fall relative to the world price.

Taking a general equilibrium perspective, however, the impact of a widespread increase in openness on relative prices in particular sectors will depend on their performance relative to other sectors. Only those sectors with productivity gains relative to the economy average would be expected to exhibit improvements in their relative price performance.[7] To the extent that the manufacturing sector is small relative to the rest of the economy, however, and provided productivity improvements resulting from internationalisation are concentrated in, and fairly evenly spread across manufacturing, one could expect to find relative price falls across much of the manufacturing sector as a result of internationalisation.

3.2 Changes in Relative Prices across the Manufacturing Sector

Changes in relative prices within different manufacturing industries are examined using data from the United Nations Industrial Development Organisation (UNIDO) database and producer price data from the Australian Bureau of Statistics. The UNIDO database covers a large number of industrialised and developing countries, and, for each country, contains annual data on nominal and real output for up to 30 manufacturing industries over the period from the late 1960s to the mid 1990s. Producer price indices are constructed from the UNIDO database by taking the ratio of nominal to real output for each industry in each country.

The Australian producer price data are available for 27 manufacturing industries, and are constructed by splicing UNIDO data for the period 1969 to 1976 to ABS producer price data for the period 1976 to 1994.[8]

The foreign price indices are constructed using price movements in 16 of Australia's major trading partners. In order to capture the competitive pressures placed on domestic producers as closely as possible, foreign prices are chain-weighted together using a moving average of foreign country export shares to Australia for each industry.[9] The foreign prices are converted to Australian dollars using bilateral exchange rates, and are tariff adjusted using Industry Commission estimates of average duty rates.[10] Transportation costs are not adjusted for in this analysis.

The resulting ratios of domestic to foreign output prices are shown in Figure 1. Summary statistics regarding movements in these indices are presented in Table 1. In drawing conclusions from these data, we have kept in mind the fact that international comparisons are fraught with difficulty. Even in industrialised countries, as suggested by the Boskin (1996) report, it is difficult to accurately measure price movements and appropriately adjust for quality and compositional changes. Many of these problems are likely to be exacerbated in developing countries. When examining these indices we are thus sceptical about the magnitude of some of the individual movements, and take from these data some overall trends rather than placing too much weight on individual industry experiences.[11] Note that since these indices have been calculated using price indices rather than actual price levels, only conclusions regarding changes in relative prices can be drawn.[12]

Figure 1: Relative Prices
Figure 1: Relative Prices
Table 1: Changes in Relative Prices by Industry
Number of industries for
which relative prices have:
1969/72 to
1991/94
1969/72 to
1980/83
1980/83 to
1991/94
Decreased 21 11 21
Increased 1 4 1
Unchanged 5 12 5
Note: Based on percentage changes in the relative price measures presented in Figure 1. Industries are classified as ‘unchanged’ if the relative price change is less than 10 per cent in either direction.

The main conclusion we draw from these data is that there have been widespread and substantial falls in relative prices for a large number of manufacturing industries over the three decades considered. Table 1 illustrates that over the full sample period, relative prices fell in 21 of the 27 industries, and only rose in one industry.[13] We discuss this conclusion in more detail below.

These data also illustrate that relative price shocks can be quite large and persistent, driving relative prices away from long-term trends. Following the nominal exchange rate depreciation in 1986, for example, most industries experienced a fall in their real exchange rate (Figure 1). For some industries this gain was subsequently eroded, while for others it was maintained.[14] We attribute the persistence of relative price shocks to rigidities in the economy which imply that it takes time for consumers and producers to observe and respond to relative price changes.[15] One implication of this is that once-off movements in the price level which flow from structural adjustments, may take place over quite a long period as relative prices gradually adjust.

Returning to the main conclusion, the widespread fall in relative prices illustrated in Figure 1 and Table 1 could reflect a number of factors.[16] These include:

  1. increasing international competition;
  2. the depreciation of the aggregate real exchange rate;
  3. real unit labour cost increases in the manufacturing sectors of our trading partners;
  4. differentiated product pricing behaviour; and
  5. the fact that much of Australia's manufacturing involves refinement of resource commodities.

We discuss each of these in turn.

The first explanation for the widespread improvement fall in relative prices is that it reflects the impact of increased international exposure on the behaviour of domestic producers. To the extent that Australian firms were, in the past, somewhat protected from international competition, through both explicit trade policy and an inward-focused mentality, it seems plausible that increased international exposure may have lead to some rationalisation in the domestic industry. This rationalisation could plausibly include lower margins in factor, input and product markets, improved use of technology by domestic industry and the exiting from the industry of high-cost producers.

The most obvious competing explanation for the observed trends, however, is that they simply reflect, at an industry level, the depreciation of the real exchange rate at an aggregate level which occurred over this period. Since one could argue that the aggregate depreciation has been driven by the decline in the terms of trade, these developments could have been entirely independent of the process of internationalisation. If this is the case, the correct interpretation would be that developments at an aggregate level had simply conferred on manufacturers substantial competitive improvements.

A third explanation is that these falls in relative prices could have been driven by general equilibrium forces at work in our major trading partners, rather than domestic developments. Taking the textiles, clothing and footwear industries as an example, the substantial relative price improvements illustrated in Figure 1 must, in part, reflect the fact that wage rates have risen substantially in some of our Asian trading partners as a consequence of development.[17]

A fourth explanation for these developments comes from thinking about the output of manufacturers as being differentiated rather than homogeneous products. In this case the law of one price is no longer strictly applicable, and the widespread fall in the real exchange rate facing most Australian manufacturing industries could reflect their attempt to gain market share.

Finally, these developments could be driven by the output of Australian manufactured industries being, on average, less elaborately transformed than their foreign counterparts. Lattimore (1988) argued that at an aggregate level, Australian manufacturing output is dominated by the simple transformation of resource-based products, whose prices have been falling over time due to sluggish world demand and growing world supply of these products. While this may be less true these days, to the extent that the output of individual Australian manufacturing industries remains, on average, more resource-based than our foreign competitors, the same argument applies. In this case the downward trend in relative prices would not reflect an improvement in competitiveness at all.

There is little doubt that the latter three explanations are relevant for understanding the observed developments. Increases in wage rates in some of our trading partners can no doubt partially explain developments in the ‘light’ manufacturing industries such as the textile, clothing and footwear industries. That manufactured goods are differentiated can probably explain why relative prices across countries can deviate over short- to medium-term horizons. And, for a selection of these industries it is likely that goods produced in Australia are more resource-based than goods produced in other countries. Overall, however, it seems unlikely that these explanations are sufficient to explain the fairly widespread trends observed across the manufacturing sector over the past 25 years.

To assess the relative importance of the first two factors, we examine their differential implications for trade shares and compare these with actual developments. Assuming domestic industries have some market power,[18] having an exogenous depreciation of the real exchange rate is equivalent to conferring on domestic tradeable industries an outward shift in the demand curve they face. This is because the increase in foreign prices relative to domestic prices (in domestic currency terms) should lead consumers to switch their demand from foreign to domestically sourced goods. Across industries, falls in relative prices would thus be expected to be correlated with reductions in import shares.

Increased international competition, by contrast, drives a fall in the real exchange rate facing the domestic industry by forcing a reduction in the price at which goods are sold in the domestic market place. For domestic producers to survive, they are forced to either lower margins and/or improve their use of available technology. High-cost domestic producers who cannot restructure their operations are priced out of the market and exit from the industry. The initial impact of an increase in international competition is thus to increase the import share, as high-cost domestic products are replaced with imports. To the extent that the restructured domestic industry subsequently improves its competitive position vis-à-vis foreign producers, the increased import share could be partially unwound.

Figure 2 shows scatter plots, by industry, of changes in relative prices against average levels of, and changes in, import shares. These show that at an industry level there is very little evidence that changes in relative prices and import shares are positively related; in only 3 of the 27 industries are changes in relative prices and changes in import share of the same sign. Combined with the fact that the domestic import-competing sector has, if anything, been declining over time, these results suggest that it is not simply an exogenous depreciation of the exchange rate which has caused these changes in relative prices.

Figure 2: Relative Prices and Openness by Industry
1969 to 1994
Figure 2: Relative Prices and Openness by Industry

While Figure 2 is suggestive of a negative association between relative price changes and openness, these data do not show that industries which have opened up the most have enjoyed significantly larger falls in relative prices. Even after controlling for the relative levels of concentration in the industry, and allowing concentration to interact with trade exposure, a significant negative correlation between relative price movements and trade exposure could not be found.[19] This may suggest that to uncover the cross-industry relationship between relative price changes and trade exposure it is necessary to control for other product-specific factors and market-structure indicators, including those additional factors discussed above.[20]

3.3 Changes in Price Setting Behaviour over Time

The above analysis suggests that there has been a widespread fall in Australian prices relative to foreign prices across the tradeable industries since the late 1960s. In this section, we provide a preliminary investigation of whether a structural change in price setting behaviour has occurred with the opening up of the economy. In particular, we examine whether across the manufacturing sector there has been a greater tendency for domestic prices to move with foreign prices in the later years of our sample, as would be expected if internationalisation has changed the price setting behaviour of domestic producers.[21]

To explore this issue, we examine whether the price ratio series we derived in Equation (1) are stationary. Identifying stationarity in a series which is subject to substantial and persistent relative price shocks, however, can be difficult. This is especially the case in small samples, where short-term relative price shocks may obscure longer-term trends. To partially overcome these problems, we follow the approach adopted in Wu (1996) and take advantage of the panel-data set to increase the power of the tests for stationarity. This approach involves pooling the price ratio series for the 27 industries, and testing the null hypothesis that each series contains a unit root against the alternative that the series are jointly stationary.

We assume a fixed-effects model, and allow the intercept to vary across industries to incorporate industry-specific effects. Our model is thus:

where Ri,t is the price ratio, defined in Equation (1), of industry i in period t, αi is the industry-specific effect, and ui,t the idiosyncratic disturbance.[22] Under the null hypothesis that the relative price series are non-stationary, the coefficient ? will equal unity. If this is rejected, and ? is significantly less than unity, it suggests that the relative price terms are jointly stationary and hence that domestic and foreign prices are linked together over time.

The results are presented in Table 2. They show that over the full sample period, the estimate of ? is very high, at 0.85, suggesting that shocks to the relative price series are long-lived. The test statistics imply that the null hypothesis of non-stationarity cannot be rejected, and hence, that relative prices have not exhibited mean reversion over the full sample, a result which is consistent with casual observation of Figure 1.

Table 2: Augmented Dickey-Fuller Tests
1969–1994 1969–1978 1978–1986 1986–1994
Estimate of ? 0.85 0.40 0.35 0.15
(0.02) (0.09) (0.13) (0.09)
[0.65] [0.45] [0.96] [0.00]
Notes: Standard errors are presented in parentheses ( ), p-values are in brackets [ ].
The finite-sample distributions of the test statistics for ? are estimated using Monte Carlo simulations calibrated to the sample.[23]

Splitting the sample into three sub-periods, these estimates suggest that domestic prices have been more closely aligned with foreign prices in recent years than earlier. The second last column of Table 2 shows that relative prices were most unstable in the middle sample period, when non-stationarity of the relative price series can be rejected at all conventional significance levels. For the sample period from 1986 to 1994, by contrast, the estimate of ? is relatively low at around 0.15, and critical values imply that the null hypothesis of non-stationarity can be rejected at a 1 per cent significance level.[24]

These results thus provide some econometric evidence that there was a realignment of relative prices in Australia during the 1970s and 1980s, and that for the past decade or so, there has been more of a tendency for domestic prices to respond to foreign prices.[25] The latter issue is investigated further in Section 4.

Footnotes

Although the existence of differentiated products and rigidities implies that the law of one price will not hold exactly, as long as foreign goods are reasonable substitutes for domestically produced goods (and as long as markets are reasonably open) the law of one price should provide a guide to longer-term relative price developments. [6]

This is a version of the Belassa-Samuelson effect. Imagine a country produces differentiated products in two traded goods industries, say footwear and computers, and opening up to trade primarily increases technological growth (and thus efficiency) in the computer industry. This will tend to increase the economy-wide wage rate, increase prices and worsen the country's competitiveness in the footwear industry. Hence, from a general equilibrium perspective, opening up to foreign competition will not necessarily improve a country's competitiveness in each of its traded goods sectors. [7]

The ABS producer price data provide a better indicator of price developments in many industries, since the UNIDO industrial production indices are, in many cases, based on more aggregated industry groupings. See Appendix C for further discussion of this issue. Overall conclusions, however, are not substantially affected by use of the UNIDO data throughout. [8]

An alternative set of weights, using shares in total world production or shares in total world trade by industry, would capture both actual and potential sources of competition. These, however, were thought to less accurately reflect the competitive pressures facing domestic producers. In particular, transportation costs (which we do not control for) may offset the competitive advantages held by some potential foreign suppliers. [9]

These measure the extent to which import prices are raised by tariffs, tender premiums on quotas and excise. For further details about the data sources and series construction, see Appendix C. [10]

In Appendix A, we present data which are comparable to Figure 1 and Table 1, but only include Australia's industrialised country trading partners. Overall conclusions are little changed. [11]

Note also that in this exercise we are comparing movements in the Australian price with the movements in the price at which a foreign substitute could be landed, not movements in the price at which foreign goods are actually sold, in Australia. [12]

Bloch (1996) uses a similar data set, but does not draw this conclusion. He finds that ‘…the change in competitiveness of Australian producers differs substantially across industries and trading partners’. For each industry, Bloch documents the number of countries against which Australia's competitiveness has improved or declined over the period from 1968 to 1989 and presents the average change in competitiveness for each industry, in effect giving equal weight to each trading partner. The differences between our conclusions primarily illustrate the effect of weighting foreign country prices together, rather than examining individual bilateral movements. In this way, more general conclusions regarding Australia's competitive position can be drawn. [13]

The differential responses of industries to the exchange rate movements in the mid to late 1980s broadly correspond with Lattimore's (1988) findings. [14]

Evidence is well established that exchange rate movements are not passed through to wholesale import prices instantaneously; see Dwyer, Kent and Pease (1993) for a discussion of some aggregate results for Australia. This is clearly an important rigidity which partially explains the persistence in relative price shocks over short periods of time. [15]

A change in transportation costs is a potentially important explanator for these trends which we have not explored. [16]

If the prices of men's shirts in Australia, the US and Hong Kong are compared since the early 1980s, Australian prices have fallen substantially relative to Hong Kong prices, but only marginally so when compared with US prices. While tariff reductions may have forced the exit of high-cost firms from the Australian clothing industry during the 1980s, the predominant reason why Australian prices have improved relative to Hong Kong's is probably because rapid technological growth in Hong Kong's high-technology traded industries has driven up the aggregate wage rate and hence, worsened their competitiveness in sectors such as clothing, which are less technology intensive and more labour intensive. [17]

As long as domestically produced goods are not perfect substitutes for foreign produced goods, the demand curve faced by domestic industries would be expected to be downward sloping. [18]

Several studies, see for example, Bloch (1974) and Levinsohn (1993), argue that it is not trade exposure on its own, but the interaction between trade exposure and concentration in the industry which is crucial. This makes sense; if a domestic industry was highly competitive before increasing its trade exposure, the disciplinary effect of increasing its trade exposure would be expected to be limited. [19]

Controlling explicitly for changes in quantitative restrictions may also be important, since the application of quantitative restrictions, rather than tariffs, generally provides more protection for the domestic industry from foreign competitors. [20]

In effect this involves testing a necessary, but not sufficient, condition for the law of one price to hold. [21]

Rather than estimate Equation (2) with its numerous dummy variables, the industry-specific effects can be removed from the data by subtracting from each observation its industry mean. An augmented Dickey-Fuller test can then be used to test the stationarity of the transformed data by estimating the following equation:

Equation
View MathML

where Inline Equation. The equations are estimated with two lags to correct for possible serial correlation over a two-year period. Similar results are obtained if time-specific, in addition to industry-specific, effects are allowed for. [22]

For the full sample period, this involves generating 27 synthetic data series, each a random-walk time series with 26 periods. Values of t? are obtained by pooling the series and using them to estimate Equation (4). This process is repeated ten thousand times to produce the empirical distribution of the test statistic. To generate the distribution of t? for the smaller sub-periods, the same process was used but with random-walk series of 10 and 9 periods. [23]

Rolling regression estimates through the sample period, using a fixed 10 year window, show that there were periods in the early 1970s in which relative prices were also stable. This is somewhat puzzling. It is probably explained by the fact that the nominal exchange rate appreciation which occurred in 1973 was substantial, but was reversed relatively rapidly. Since domestic and foreign price levels moved similarly over this period, the real exchange rate exhibits a reasonable degree of ‘mean reversion’ due to the nature of the exchange rate shocks which occurred. [24]

We regard these results as indicative of the changes which have been taking place. However, due to the characteristics of our data set, we rely more on their qualitative, rather than their precise quantitative, information. For example, the UNIDO data are collected from individual country sources, and since not all countries collect data on a calendar year basis we cannot be confident that measurement error due to timing issues has been avoided. [25]