RDP 9309: Alternative Concepts of the Real Exchange Rate: A Reconciliation 4. Evidence of a Disparity

4.1 Measuring Real Exchange Rates

Estimates of the PPP based real exchange rate, Rp, are obtained by adjusting a weighted nominal exchange rate index for differential movements in prices between the domestic economy and foreign economies. Institutions such as the International Monetary Fund, the OECD and Morgan Guaranty publish a variety of real exchange rate indices for various countries. In this paper we use the Morgan Guaranty series for our international comparisons as they are published for a broader range of countries and for a longer time period that are the other series.[23] For Australia, we use the Reserve Bank's estimate of the real trade weighted index (TWI), as this measure of the real exchange rate is very similar to the Morgan Guaranty series. Data are described in the Appendix.

As discussed by Dwyer (1992), estimates of the domestic relative price form of the real exchange rate, Rr, can be obtained in two main ways:

– the ‘direct’ approach, in which the price of output from broad industry divisions nominated as representative of the non-traded goods sector is compared with the price of output of industries representing the traded goods sector; and

– the ‘indirect’ approach, in which the relevant output prices are not known. The relative price of non-traded goods is measured indirectly by dividing an index of the general price level by the price of traded goods, with the latter proxied by an average of export and import prices.

The direct approach relies on the availability of implicit price deflators or price indices for specific classes of goods. Whilst most examples of this approach use highly aggregated data and generate crude proxies of relative prices, an exception is that by Dwyer (1992) in which a sophisticated estimate of relative prices for Australia is presented. The results of Dwyer (1992) have been updated and will be used here to represent relative domestic prices in Australia.[24] Unfortunately, however, equivalent measures are not available for other countries. In consequence, to obtain a proxy for relative prices abroad, the indirect approach was employed as it does not require sectoral price indices. Following Pitchford (1986), the general price level is measured by the CPI and divided by a simple average of the implicit price deflators for exports and imports.[25]

4.2 Australia's Real Exchange Rate

The real TWI (Inline Equation) and an index of domestic relative prices (Inline Equation) are illustrated in Figure 1. A fall in either index implies real depreciation. It is apparent that the two measures behave quite differently. Of the two indices, Inline Equation has been considerably more volatile, with episodes of sharp depreciation in the mid 1980s and again more recently. In contrast, Inline Equation, after moving in a relatively narrow band in the 1970s, has exhibited a pronounced upwards trend over the 1980s.

Figure 1: Measures of Australia's Real Exchange Rate March 1972 = 100

Ultimately, in the absence of measurement problems, disproportionate movements between Inline Equation and Inline Equation must reflect a violation of the conditions which relate the two forms of the real exchange rate – that is, the law of one price and unchanging relative prices abroad.

Whilst it is assumed that the law of one price holds in the long run, short-run departures from the law of one price are self evident and will contribute to disparities between Inline Equation and Inline Equation. This is so because whilst changes in the nominal exchange rate impact immediately on Inline Equation, this need not be the case for Inline Equation. Unless exchange rate pass-through is both complete and instantaneous, movements in domestic relative prices will both lag those of Inline Equation and exhibit smaller oscillations. However, other things being constant, when the process of pass-through is complete, the two forms of real exchange rate should move similarly.[26]

Instead, the focus of this paper is on situations in which there may not be an automatic tendency for Inline Equation and Inline Equation to move similarly. These situations emerge when relative prices at home and abroad move differently, reflecting real factors such as changes in productivity and terms of trade shocks.

First, the secular increase in the relative domestic price index is suggestive of trend growth in the productivity of the tradeables sector exceeding that of the non-tradeables sector. As shown analytically, if the increase in the relative price of non-tradeables abroad is greater than that at home, the PPP based real exchange rate will depreciate even though the domestic relative price of non-tradeables is increasing. This appears to have happened over the 1980s: Inline Equation has increased whilst Inline Equation has fallen.

Second, the two forms of real exchange rate appear to respond differently to movements in the terms of trade. As noted in the theoretical discussion, the terms of trade should be positively correlated with Inline Equation. The strength of this direct relationship is illustrated in Figure 2.[27] It is most evident during the major swings in the terms of trade during the early 1970s and mid 1980s.

Figure 2: PPP Deviations and the Terms of Trade March 1972 = 100

Conversely, it was noted in the theoretical discussion that Inline Equation should be negatively correlated with the terms of trade when export prices change. A general inverse (although not tight) relationship between the terms of trade and Inline Equation is suggested in Figure 3. This result is consistent with observations that Australia's terms of trade shocks are most often generated by changes in export prices. For instance, in the early 1970s, the price index for non-tradeables increased more slowly than that for tradeables so that the relative price index fell. This occurred at a time when the terms of trade improved due to a rise in export prices. The same inverse relationship is apparent during the major changes in the terms of trade in the late 1970s and the mid 1980s. Thus changes in export prices are likely to produce pronounced differences between the two forms of real exchange rate, whilst differences in trend productivity growth will contribute to an increase in the underlying divergence between them.

Figure 3: Relative Domestic Prices and the Terms of Trade March 1972 = 100

4.3 International Comparisons

We now turn to a number of international comparisons, paying particular attention to two issues:

  1. Is the Australian experience of an increase in the relative price of non-tradeables common to other countries?
  2. What relationship exists between movements in the terms of trade and the real exchange rate in other countries?

4.3.1 Secular Movements in Domestic Relative Prices

In this section we present estimates of domestic relative prices (Inline Equation) for a range of countries. Figure 4 presents Inline Equation for Australia, New Zealand and Canada. Each of these countries is a small open economy that is subject to large relatively swings in its terms of trade. Figure 5 presents Inline Equation for four European countries: France, Germany, Italy and the United Kingdom. Figure 6 presents this domestic relative price index for the United States, Japan and South Korea. Figure 7 shows the index for Norway, the Philippines and Pakistan. These figures show that over the 1980s, in all countries (except for Pakistan), the price of non-traded goods has increased at a faster rate than the price of traded goods. Clearly, the Australian experience of a secular rise in the relative price of non-tradeables is not unique.

Figure 4: Relative Prices: CPI Divided by Price Index for Tradeables March 1972 = 100
Figure 5: Relative Prices: CPI Divided by Price Index for Tradeables March 1972 = 100
Figure 6: Relative Prices: CPI Divided by Price Index for Tradeables March 1972 = 100
Figure 7: Relative Prices: CPI Divided by Price Index for Tradeables March 1972 = 100

Simple proxies of domestic relative prices must be interpreted with caution. In particular, it is difficult to relate directly movements in such prices to differential rates of growth in productivity in the traded and non-traded goods sector, where similarly crude proxies are used.[28] Furthermore, it is difficult to separate the effect of productivity bias from other effects, especially when countries are characterised by different degrees of stability in their general price level or in their terms of trade. Such an exercise in beyond the scope of this paper. Nonetheless, for those countries with unambiguously high rates of productivity growth in the traded goods sector (such as Korea and Japan) the rise in the relative price of non-tradeables has been greater than elsewhere. Conversely, in those countries where productivity growth in the traded goods sector is unambiguously low (such as Pakistan and the Philippines), the rise in the relative price of non-tradeables has been less than elsewhere.

The figures reveal another important aspect of trend movements in domestic relative prices. In almost every country, the general increase in the relative price of non-traded goods was restricted to the 1980s. Of the 13 countries examined, only one experienced a net increase in the relative price of non-tradeables between 1972 and 1980. This raises the question of why the experience of the 1980s is so different to that of the 1970s? To a large extent, the answer lies in the substantial increase in oil prices in 1974 and 1979. These increases represented a direct increase in the relative price of traded goods. In addition, in oil importing countries, higher oil prices represented a sizeable shock to real income. This income shock placed further downward pressure on the relative price of non-tradeables. The reverse process occurred in 1986 when oil prices fell substantially. Lower oil prices meant lower traded goods prices. Further, in oil-importing countries the lower prices generated positive income effects and upward pressure on non-traded goods prices. Since 1986, most countries have experienced further upward pressure on the relative price of non-tradeables. The influence of the oil price shocks invites further consideration of the relationship between real exchange rates and the terms of trade.

4.3.2 Real Exchange Rates and the Terms of Trade

In the case of Australia it was argued that increases in the terms of trade are likely to lead to an appreciation of the PPP based real exchange rate, but to a decrease in the relative price of non-tradeables. This inverse relationship between the terms of trade and the relative price of non-tradeables reflects the fact that, in general, changes in export prices, rather than import prices drive the terms of trade. In contrast, in countries for which import prices are more volatile than export prices, a positive relationship between relative prices and the terms of trade should be the result. This point can be seen in the following figures. These graphs show the terms of trade, domestic relative prices and PPP based measures of the real exchange rate for four countries: Japan, Korea, the United States and Norway. For the first three of these countries, import prices are more volatile than export prices. In contrast, Norway, like Australia, experiences greater volatility in its export prices than in its import prices.

The PPP based real exchange rate presented in these graphs is the Morgan Guaranty measure (Inline Equation). Recall that this measure uses the wholesale price of non-food manufactures to deflate the nominal effective exchange rate index. This price series comprises a disproportionate share of traded goods and may not be a good measure of the general price level in all countries, especially those with a high share of non-tradeables in their consumption bundle or those in which relative price changes have been pronounced. Unfortunately, PPP based real exchange rates that include general consumer prices are not readily available back to 1972 for the countries under review. In consequence, a real exchange rate that incorporates a general consumer price index was also constructed.[29] The result was little different for all countries for which it could be calculated, with the exception of Japan. For Japan, the two measures are quite different and thus both are shown.[30]

We begin by examining the Japanese case. Figure 8 shows the Japanese terms of trade, relative domestic prices and the two measures of the PPP based real exchange rate. Japan appears to be the classic textbook case. Large improvements in its terms of trade lead to an appreciation of both PPP based measures of the real exchange rate and also to an increase in the relative price of non-traded goods, as is the traditional view.

Figure 8: The Real Exchange Rate, Relative Prices and the Terms of Trade: Japan March 1972 = 100

This outcome for Japan reflects several key points outlined in the theoretical discussion. First, the trend increase in the relative price of non-tradeables is consistent with faster productivity growth in Japan's traded goods than in its non-traded sector. Second, the appreciation of the PPP based real exchange rate is consistent with faster productivity in Japan's traded goods sector relative to those abroad. (In fact, in this regard, it is not surprising that the PPP based measure which uses consumer prices has appreciated considerably more than the Morgan Guaranty index as the former will include a greater share of prices of non-tradeables.) Third, the positive correlation between both forms of real exchange rate and the terms of trade arises from the fact that shocks to the Japanese terms of trade come primarily through shocks to import prices, in particular oil.[31]

The Korean case is a little more difficult to interpret (see Figure 9). It might be expected that a faster rate of growth in productivity of Korea's traded goods sector would drive up the relative price of non-traded goods by more than in its trading partners so that both forms of real exchange rate would appreciate. Instead, only the domestic relative price index increased. Domestic relative prices are, in general, positively correlated with the terms of trade. However, there has been a slight downward trend in the conventional real exchange rate. In fact, movements in the real value of the currency appear, to a significant extent, to be detached from the terms of trade.

Figure 9: The Real Exchange Rate, Relative Prices and the Terms of Trade: Korea March 1972 = 100

In the United States, movements in relative prices have been much less pronounced than in either Korea of Japan (see Figure 10). However, just as in Korea and Japan, the relative price of non-tradeables has been positively correlated with the terms of trade. So too has the PPP based measure of the real exchange rate, suggesting that terms of trade shocks in the United States are typically caused by changes in import prices.[32]

Figure 10: The Real Exchange Rate, Relative Prices and the Terms of Trade: USA March 1972 = 100

Finally, the Norwegian experience shown in Figure 11 is similar to that of Australia: the terms of trade is negatively correlated with the relative price of non-tradeables. Norway is, however, an oil exporting nation. Thus the improvement in Norway's terms of trade following the oil price increase in 1979 was associated with a fall in the relative price of non-tradeables. Conversely, the fall in its terms of trade following the collapse of oil prices in 1986 saw the relative price of non-tradeables increase. These changes in relative prices associated with the terms of trade have occurred against the background of a general secular increase in the relative price of non-tradeables.

Figure 11: The Real Exchange Rate, Relative Prices and the Terms of Trade: Norway March 1972 = 100

Footnotes

In the Morgan Guaranty series, weights are proportional to the absolute value of trade between the domestic economy and its trading partners. In adjusting for differences in inflation across countries, the price of non-food manufactures is used. [23]

Dwyer (1992) uses input-output tables to classify each of the 108 industries in the commodity details of the national accounts as either tradeable or non-tradeable and then constructs the relevant price indices using these classifications. The measure draws on an unpublished ABS data set that covers the period from the March quarter 1970 until the March quarter 1990. However, the series has since been updated to the June quarter 1992. In the international literature, see also Mellis (1993) for a sophisticated direct estimate of relative prices in for a number of European countries. Mellis uses detailed consumer price data to construct a relative price series for goods (tradeables) and services (non-tradeables). [24]

Estimates of relative prices based on this method need to be interpreted with some caution. Recalling that α is the share of non-traded goods in the CPI, this measure can be expressed as:

Equation

If α is different in each economy, movements in this proxy of relative prices are not directly comparable across counties. In addition, abstracting from other measurement problems, movements in this index will underestimate movements in actual relative prices as α is less than one. (This underestimation is greater the smaller is α.) [25]

As discussed in Dwyer, Kent and Pease (1993), the bulk of exchange rate pass-through to free on board import prices occurs in the first two quarters and is complete after one year. A similar rate of pass-through is found for commodity exports (Heath 1991). Thus to the extent that departures from the law of one price drive disparate movements in the two forms of real exchange rate, this is likely to be significant for only two or three quarters. [26]

The implicit price deflators used in the calculation of the terms of trade exclude computers. This is so for two main reasons. One, their prices have fallen rapidly in recent years, significantly affecting the aggregate deflator for imports. Two, the method employed by the Australian Bureau of Statistics to estimate computer prices differs to that adopted by other agencies and detracts from the comparability of domestic and foreign prices of traded goods. [27]

The essential problem is accurately identifying what constitutes the traded and non-traded goods sector in each country. Sectors are most often defined as traded or non-traded a priori rather than according to a formal classification system. Whether or not prices, output or productivity rates are representative of a given sector is rarely tested. [28]

Given that a consistent published series of PPP real exchange rates based on consumer prices is not available back to 1972, a simple proxy was calculated. The Morgan Guaranty nominal trade weighted index (against industrial countries) was multiplied by the home country CPI and then divided by the IMF's estimate of an aggregate industrial country CPI. [29]

Similar differences might be expected for Korea. However, this cannot be tested readily as Morgan Guaranty do not estimate a nominal effective exchange rate for Korea which can be adjusted for differences in consumer prices domestically and abroad. [30]

A qualification is required here. The Japanese terms of trade is, to some extent, endogenous: by improving productivity in its traded goods sector, it reduces the relative price that it receives for its exports, leading to upward pressure on its real exchange rate. However, major swings in the terms of trade are exogenous and tend to come about through increases in the world price of its inputs. In this case, both forms of the real exchange rate are positively correlated with the terms of trade, in accordance with conventional wisdom. [31]

In the United States, the terms of trade are, to a significant extent, endogenous. For this reason, one might expect positive correlation between the PPP based real exchange rate and the terms of trade. However, the US is subject to some exogenous shifts in its terms of trade, in particular due to oil price shocks. Of interest is the fact that, at these times, the relative price of non-tradeables also exhibits positive correlation with the terms of trade, consistent with the proposition that terms of trade shocks are sourced to import prices. [32]