Accounting for Real Exchange Rate
Changes in East Asia
David C. Parsley
Owen Graduate School of Management
Vanderbilt University
Nashville, TN 37203
[email protected]
First version: May 2001
This version: February 2003
Abstract
This study measures the proportion of real exchange rate movements that can be accounted
for by movements in the relative price of non-traded goods among twenty-one bilateral AsianPacific real exchange rates. Following Engel (1999), the decomposition is done at all possible
horizons that the data allow – from one month up to 25 years. For the most part, evidence
presented here is consistent with that from his sample of (predominately) G7 countries. In
particular, relative prices of non-traded goods appear to account for virtually none of the
movements in Pacific Rim real exchange rates. This pattern appears unaffected by the crosssectional variation in either income level, or the degree of openness present among these
Pacific-Rim economies. The exceptions to these results occur when we examine the drift in
real exchange rates, and more generally, for fixed (or semi-fixed) exchange rate regimes.
_________________
The first draft of this paper was completed while I was a visiting research fellow at the Hong
Kong Institute of Monetary Research. The views expressed in the paper are mine and should
not be attributed to the HKIMR or to its Board of Directors. I thank Michael Devereux,
Stefan Gerlach, Vikas Kakkar, Guy Meredith, and seminar participants at the HKIMR, and at
City University of Hong Kong for helpful comments.
1
1.
Introduction
This study provides new evidence on the share of real exchange rate movements that
can be accounted for by deviations from purchasing power parity in traded goods. The
primary motivation for this exercise is the influential evidence presented in Engel (1999). For
a sample of high-income countries, he showed that over the past thirty years, movements in
relative prices of non-traded goods appear to account for essentially none of the movements
in U.S. real exchange rates.
That is, movements in aggregate real exchange rates
overwhelmingly reflect deviations from purchasing power parity (PPP) among traded goods.
This finding is important, since many general approaches to modelling the real exchange rate
are built on an assumption of PPP in traded goods. Indeed, as emphasized by Obstfeld and
Rogoff (2000), a role for relative traded goods prices in real exchange rate determination is
absent in most standard theories. The evidence presented in this study focuses on the
experience of six small, open, and generally fast growing Asian-Pacific economies.
Despite the striking nature of Engel’s findings, there are several a priori reasons one
might question their generality.
First, the focus on larger, predominately western G-7
countries misses much of the impressive economic growth that has occurred over the last
twenty-five years. It is possible that wide movements in the relative price of non-traded goods
have accompanied these high rates of growth, and that the accompanying sectoral
reallocations have influenced the behaviour of real exchange rates. Second, the exchange rate
regime may itself be important. Recent studies have found significant effects on trade,
growth, and economic integration (Rose 2000, Frankel and Rose 2002, and Parsley and Wei
2002). Several of the economies included in this study pegged their exchange rates for
extended periods (e.g., Hong Kong 1983-present, Thailand 1990-97), thus precluding nominal
exchange rate adjustment. To the extent that the nominal exchange rate regime affects price
dynamics, the real exchange rate may also be affected. Third, the Asian-Pacific economies are
generally more reliant on external trade than the high-income industrial countries often
studied. This openness may also affect real exchange rate dynamics. Finally, there are wide
disparities in income levels among the Asian-Pacific economies themselves.
Thus, an
2
interesting question is whether the benchmark presented in Engel (1999) is relevant to a set of
countries with widely different economic circumstances.
Recent studies that have focused on the importance of relative non-traded goods
prices in real exchange rate determination have had mixed results. Chinn (1996) finds some
East Asian real exchange rates (Japan, Korea, the Philippines, and Singapore) do appear
cointegrated with relative prices, while others (China, Indonesia, and Thailand) do not. For
OECD countries, Canzoneri, Cumby and Diba (1999) find mixed evidence for cointegration
between real exchange rates and relative prices. Kakkar and Ogaki (1999) also find mixed
results for the U.S., Japan, the U.K., and Canada. More recently, Chinn and Johnston (1999)
and Strauss (1999) find somewhat stronger support for this linkage among OECD countries.
This paper follows Engel (1999) and ‘accounts’ for the share of real exchange rate
movements attributable to deviations from the PPP in traded goods. The accounting exercise
is based on a straightforward application of the definition of aggregate price indexes. To be
specific, express the (log) aggregate price index as a weighted-average of traded (T), and nontraded (N) goods prices, i.e.,
p = (1 − α ) p T + α p N ,
for the domestic country and,
p = (1 − β ) p
for the foreign country.
*
T*
+β p
N*
Then the real exchange rate, q = s + p * − p , may be written as the sum of the relative price of
traded goods (x), and the relative-relative price of non-traded goods (y).
q = s + pT * − pT + β ( p N * − pT * )− α ( p N − pT ) ≡ x + y
(1)
where : x = s + p T * − p T , and y = β ( p N * − p T * ) − α ( p N − p T ) .
This decomposition illustrates how differences across countries in internal relative
prices impact the real exchange rate even in the absence of nominal exchange rate changes.
As a concrete example, the real appreciation of the yen over the 1975-1995 period is generally
associated with the relatively faster output growth in Japan (e.g., Kakkar and Ogaki (1999), or
Krugman and Obstfeld, 1999). However, a key implication of the decomposition described
3
by equation 1 is that the domestic relative price of non-traded goods is only part of y. For the
domestic relative price of non-traded goods to have an impact on real exchange rate
movements, the foreign relative price of non-traded goods must not move similarly. The
primary objective of the accounting exercise is to apportion variation in the real exchange rate
into shares attributed to variation in x, and variation in y, separately.
Before presenting the decompositions, Table 1 presents some summary data
comparing the countries examined in this study to those studied by Engel (1999). In the first
column annual average real GDP growth since 1975 is tabulated. The economies in the upper
panel have (with the exception of Malaysia) generally doubled the growth performance of the
higher income economies in Engel’s sample. In the second column note that income levels,
as measured by real per-capita GDP, are on average barely half those in Engel’s study. Finally,
openness – as measured by the sum of exports and imports relative to GDP – is given in
column 3.
As noted above, the Asian-Pacific economies are generally more open.
In
principle at least, these differences in economic circumstances may have contributed to very
different roles for deviations from the law of one price, and relative shifts in internal relative
prices (i.e., x and y shares) in the real exchange rate movements of these East Asian
economies. It would however, be misleading to claim these implicit criticisms are novel –
indeed, Engel discusses them briefly in his conclusion.
The approach in this paper follows Engel (1999) closely. The proportion of real
exchange rate movements that can be accounted for by movements in the relative price of
non-traded goods is computed for each bilateral currency pair, and the decomposition is done
at all possible horizons that the data allow – from one month up to 25 years. Perhaps
surprisingly, the evidence presented here is broadly similar to that obtained from high-income
U.S. based real exchange rates. In particular, relative prices of non-traded goods appear to
account for virtually none of the movement of Pacific Rim real exchange rates. This pattern
appears unaffected by the cross-sectional variation in either income level, or the degree of
openness present among these Pacific-Rim economies. The only exception to these findings
occurs when we examine fixed (or semi-fixed) exchange rate regimes separately, and when we
focus on the drift in real exchange rates (as opposed to measures of variance or mean squared
4
error).
However, the primary exception (the Hong Kong dollar) appears anomalous
compared to other managed currencies within the sample, and, as will be discussed below, to
the recent findings of Mendoza (2000).
2.
Traded and Non-Traded Price Indexes
The first step in the decomposition of real exchange rate movements is constructing
traded and non-traded price indexes. This study uses consumer price index data from six
Asia-Pacific countries: Hong Kong, Korea, Malaysia, Singapore, Taiwan, Thailand; and the
United States, to construct measures of traded and non-traded goods prices for each country.1
One advantage of obtaining data from higher income countries is that there is more data, and
often of better quality. Hence, Engel (1999) was able to construct (and decompose) five
separate measures of real exchange rates – using consumer prices, producer prices, etc. His
findings were virtually identical across all measures. Additionally, in a related exercise, Isard
and Symansky (1996) construct three annual measures of real exchange rates for each of the
economies studied here. Again, these authors find their conclusions robust to the alternative
measures. Thus in this study we focus on consumer prices exclusively.
As in Engel (1999), geometric weights for major components of the “all items” series
were estimated from logarithmic regressions of the form
(
)
∆(cpi − rent ) = ∑ φi ∆(traded i − rent ) + ∑ φ j ∆ services j − rent + ε ,
i
j
where, cpi = cpi – all items, rent = cpi – rent, etc., and the geometric weight for rent is
computed residually as:
1 − ∑ φ i − ∑ φ j .2
i
For each currency, the number of separate
j
components of traded goods, and non-traded goods (generally services) varies (between one
and three) according to the data available.
Then the traded and non-traded price indexes are computed as
The choice of countries was dictated by data availability. The countries studied are all of the countries available
in the CEIC database provided by the Hong Kong Institute for Monetary Research.
2 Appendix 1 lists the CPI sub-indexes used and the range of available data for each economy examined.
Nominal exchange rates as well as CPI data are taken from the CEIC database.
1
5
pT =
pN =
1
1−α
∑φ
1
1− β
∑φ
traded i and,
i
i
j
services j .
j
Where 1 − β = ∑ φ j services j and, 1 − α = ∑ φi traded i .
j
i
In general, traded goods price indexes consist of each country’s aggregate “goods”
components of the CPI, while the non-traded goods price indexes consist of the services and
rent components. Wherever possible, the non-traded price indexes exclude obvious traded
goods components, e.g., rent, excluding fuel and power. In the end, these remain imperfect
measures. However, Engel (1999) suggests that the effects of such miss-measurement are not
serious for accounting purposes. This conclusion is based on several robustness checks.
First, he presents results using five different measures of traded- and non-traded goods prices.
Second, he examines one case in detail (Japan), for which data on marketing services is
available. And finally, he explicitly examines the size of the bias on his estimates of α and β .
In short, his conclusions remain the same in all cases.
The next step is to construct the real exchange rates and the traded, and non-traded
components, x and y (in equation 1). In this study, it is possible to construct twenty-one real
exchange rates – six bilateral U.S. real exchange rates plus fifteen cross-rates. Appendix
figures 1 and 2 plot the nominal and real exchange rates separately. Two aspects of appendix
figure 1 are notable. First, the reaction to the recent currency crisis (1997-98) was varied, with
some nominal exchange rates (especially the HK dollar/US dollar) remaining relatively stable.
And related, the response does not merely depend on the particular institutional exchange rate
arrangement. Appendix figure 2 reveals some cases of extended trend movements as well as
the relatively large real exchange rate movements associated with that crisis.
The focus of this study is on accounting for the movements in the real exchange rates
depicted in appendix figure 2. In particular, we wish to attribute shares of the historical
movement in real exchange rates to movements in traded goods prices (x), and non-traded
goods prices (y). It is to this that we now turn.
6
3.
Accounting for Real Exchange Rates
Movements in the real exchange rate q may be due to movements in (i) the trend
(drift), (ii) the variance, or comprehensively, (iii) the mean-squared error. In each case it is
possible to decompose movements into parts attributable to traded goods (x, in equation 1),
and to non-traded goods (y, in equation 1). First consider the drift.
a. The drift in the real exchange rate
The average one-month drift in the real exchange rate is equal to
∆q =
1 N

 ∑ (qt − qt −1 ) . Then, the average n-month drift is simply n∆q .
N − 1  t =2

The time dimension is relevant since some theories may relate to the ‘short-run’, while
others concern longer run movements. In this study, all possible horizons (i.e., 1 month to nmonths) that the data allow are considered. For the data examined here, the maximum value
of n varies by country-pair. For the bilateral South Korea-U.S. real exchange rate, the longest
n-period drift is 303 months, i.e., slightly more than 25 years. For country pairs involving
Malaysia and Thailand there are considerably fewer observations, only 123 months.
Since the drift may be positive or negative, and thus potentially offsetting, we focus on
the mean squared drift, n 2 ∆q 2 . The question we are interested in is how much of the squared
drift in the real exchange rate is attributable to x. For this, we must decide how to treat comovements in x and y. For his data, Engel argued that the co-movements were small. This
turns out to be the case for these data as well. A formal test was conducted by computing the
ratio of the covariance of first differences in x to its variance, and testing whether that ratio
was statistically different from zero. The t-statistic from the test was 1.44, not statistically
different from zero at usual significance levels. Hence in subsequent calculations we ignore
the covariance term. The drift decomposition can thus be written as:
Traded goods share of n-period drift =
n 2 ∆x 2
.
n 2 ∆x 2 + n 2 ∆y 2
(2)
7
Table 2 presents this drift decomposition for each of the 21 bilateral real exchange
rates. First, note that the fraction of the drift attributable to the traded goods component
varies considerably. In thirteen out of the twenty-one cases, drift in the relative price of
traded goods accounts for virtually all (more than 80%) of the drift in the real exchange rate.
This is striking and essentially reproduces Engel’s results. However, there are exceptions.
First, the two anomalous bilateral U.S. cases – Hong Kong, and Singapore – are
simultaneously the most open, richest, and among the fastest growing. Beyond this it is
difficult to make generalizations – except noting that the importance of deviations from PPP
varies widely. In the cases of the HK dollar/Thailand baht and Malaysian ringgit/Singapore
dollar, variation in y appear very important indeed.
For most of the bilateral pairs studied, there is substantial evidence that drift in relative
traded goods prices contributes significantly to the drift in real exchange rates we have
observed. This implies substantial market segmentation among the economies studied –
despite the relative importance of imports and exports in these economies. However, for
about one-third of the bilateral real exchange rates we study, movements in the drift in nontraded goods prices has contributed as much or more to the overall drift in real exchange rates
in the region as has the drift in relative traded goods prices. This evidence is new, however
prior to interpretation we need a measure of how much of the total variation in exchange rates
can be attributed to the drift. In order to address this we must first consider the second
component of the variation in real exchange rates. We come back to the issue of the
contribution of drift to total variation in section 3d below.
b. The Variance of the real exchange rate
A second aspect of real exchange rate movement is its variance. We can consider the
fraction of the n-period variance accounted for by x. As discussed by Engel (1999), the nperiod sample variance has extreme small-sample bias for large n. Thus we follow Engel and
adopt the (small sample, unbiased) measure proposed by Cochrane (1988)3:
3
The covariances discussed above were also calculated using this small sample correction.
8
var( qt − q t −n ) =
(
N −n
N
q j + n − q j − n∆ q
(N − n − 1)(N − n ) ∑
j =1
)
2
Then, the fraction of the var( qt − qt − n ) accounted for by var( x t − x t −n ) is:
Traded goods share of n-period variance =
var( x t − x t −n )
var( x t − x t −n ) + var( y t − y t −n )
(3)
Figure 1 presents these decompositions graphically for all twenty-one real exchange
rates. Note that the horizontal axis measures the horizon. Here the message is stronger. In
virtually all cases, at all horizons variation in traded goods prices accounts for essentially 100%
of the variation in the real exchange rate. This is precisely the result Engel (1999) found for
his industrial country sample. The only movement (in the panels of Figure 1) away from
100% are at the very long horizons. These longer horizons should be interpreted with
caution. First, very few observations are used in the calculations and thus, the potential for
single influential observations becomes greater. Second, at these longer horizons, the impact
of the 1997-98 crisis becomes more important, since it is included in more and more of the
sample data points. Hence, to the extent the 1997-98 crisis represented an ‘unusual’ event we
might wish to put less weight on it. Finally, even in the cases with the fewest observations, the
message is that at horizons out to eight years essentially all of the variation in the real
exchange rate comes from variation in traded goods prices. We next turn to a comprehensive
measure of real exchange rate variation.
c. The Mean Squared Error of the Real Exchange Rate
The mean squared error is the sum of drift and variance components. Hence the
decomposition of the mean squared error (MSE) is:
Traded goods share of n-period MSE =
MSE( x t − x t −n )
MSE( x t − x t −n ) + MSE( y t − y t −n )
(4)
Figure 2 presents these decompositions. The message from figure 2 is exactly the
same as from the variance decompositions, i.e., virtually 100% of the total variation (MSE) of
the twenty-one real exchange rates we consider is contributed by variation in traded goods
9
prices. This is true at all horizons the data allow, from one month to over twenty-five years.
The Malaysian ringgit represents a possible exception to the above generalizations. That is at
horizons above 100 months, the importance of variation in x drops off precipitously relative
to total variation in q. This is somewhat anomalous, and undoubtedly reflects the effect of the
crisis period. Finally, and more generally, note the absence of apparent systematic with the
cross-sectional variation in openness, income level, or (nominal) exchange rate regime.
These results should come as a surprise. If PPP held, x should be a constant and
therefore contribute nothing to an explanation of variations in the real exchange rate. A
primary motivation for examining the East Asian currencies in this study was an apparent
difference in the economic conditions and experiences of these economies as compared to the
experience of high-income, moderately growing economies previously examined. Yet we
conclude that even in this special case, the data offer very little support for any of the theories
of real exchange rate behavior that rely on variation in relative non-traded goods prices – at
least as they apply to the time horizons considered here. We make several further attempts at
reconciling these results before concluding.
d. Reconciliation
An outstanding issue discussed above relates to the contribution of the drift in overall
real exchange rate variation. According to Table 2, a substantial part of the drift in the real
exchange rate could, for a third of the real exchange rates examined, be attributed to drift in
relative non-traded goods prices. Yet, according to Figure 2, virtually 100% of the total
variation in the real exchange rate (as measured by the MSE) was due to variation in relative
traded goods prices. For these two observations to be true, it must be the case that the drift
accounts for a relatively small part of the total variation in real exchange rates. Table 3
presents one relevant summary statistic: the average drift contribution to the MSE for each
exchange rate. The data support the conjecture – the maximum contribution of the drift is
18% for the Korean won/U.S. dollar real exchange rate, and except for two other cases, this
contribution is always in single digits. Hence, there is no inconsistency between Tables 2 and
3. Nonetheless, the results in Table 2 present a quite different picture than those in Engel
10
(1999). Unfortunately, it is not obvious what institutional or structural features may be
responsible for the differences in the traded goods component highlighted there.
Another issue noted in the introduction is the role of fixed exchange rates. Under
flexible exchange rates, the variability of the exchange rate could swamp variability in the two
national currency traded goods price indexes. Thus, the dominance of the traded goods
component of the real exchange rate would be driven by variability in the nominal exchange
rate. In fixed exchange rate regimes, this cannot be the case. Hence, it would indeed be
surprising if traded goods prices were necessarily more variable, and thus were always the
dominant driver of real exchange rates.
To isolate the impact of fixed (or nearly fixed) exchange rates, we look at the three
cases of fixed, or semi-fixed exchange rates in our sample, i.e., the Hong Kong dollar, the
Malaysian ringgit, and the Thai baht. In particular, we focus on the Hong Kong dollar from
1983:10 to the present, on the Malaysian ringgit from 1990:01 to 1997:05, and on the Thai
baht from 1990:01 to 1997:06. Only the Hong Kong dollar maintained a true peg; the ringgit
and baht varied within fairly narrow bounds. The top panel of Figure 3 presents the variance
decompositions (i.e., the fraction of the variance in the real exchange rate accounted for by
variance in traded goods). The bottom panel plots the MSE decompositions (i.e., the fraction
of the total variation in the real exchange rate accounted for by MSE of traded goods).
There is another reason to examine the fixed exchange rate cases separately. In a
related study, Mendoza (2000) finds that variability in the relative price of non-tradable goods
account for up to 70 percent of the variability of the peso-dollar real exchange rate during
periods in which Mexico had a managed exchange rate regime. Figure 3 presents the most
evidence yet for the importance of variation in non-traded goods prices in real exchange rate
movements. But even here, the evidence is not unanimous. For the Hong Kong dollar,
variation in the relative price of non-traded goods approaches fifty percent (of the MSE) at
six-year horizons, but at longer horizons (to seventeen years) declines back to twenty-five
percent. However, note that even in this case, non-traded goods prices never approach the
seventy-percent found by Mendoza (2000), and the maximum effect for the Hong Kong
dollar occurs at much longer horizons than for the peso (six years versus six months).
11
Moreover the U-shaped pattern exhibited by the Hong Kong dollar in Figure 3 is not matched
by either the Thai or Malaysian currencies. Finally, for the ringgit, at horizons exceeding eight
years the importance of variation in traded goods prices (x) declines to roughly seventy-five
percent, but for the baht there is only a slight decline. Thus, Mendoza’s results regarding the
importance of managed exchange rates do not readily generalize, suggesting that fixing the
nominal exchange rate may be necessary, but by no means sufficient for reducing the
importance of variation in traded goods prices in real exchange rate variation.
A final question is what has happened to the components of the relative-relative price,
y. That is, it is widely believed that some economies in the region have witnessed wide relative
price changes. Why haven’t these translated into real change rate changes? The answer is
provided by the definition of y. As noted above, changes in the relative price of non-traded
goods across countries must not be ‘similar’ if such movements are to be reflected in the real
exchange rate. Appendix figure 3 presents graphs of α ( p N − p T ) and β ( p N * − p T * ) (in
levels) for all twenty-one bilateral pairs. It appears that α ( p N − p T ) and β ( p N * − p T * ) often
trend similarly, which would imply that their difference tends to zero. If true, the contribution
of internal relative prices would be less important in the overall decomposition.
More formally, it is possible to test whether changes in these internal relative price
ratios are comparable across countries.
Table 4 presents the p-values and number of
observations used to compute the test statistic, from such tests for three horizons: 1 month, 6
months, and 1 year. At the one month horizon, the only cases where internal relative prices
move dissimilarly are bilateral pairs involving Hong Kong. That is, the difference in relative
price movements is statistically different from zero (at the 5% level) only for the U.S., Korea,
Malaysia, and Thailand relative to Hong Kong. In the remaining seventeen bilateral cases we
cannot reject that the difference in relative price movements is zero. Interestingly, for these
bilateral pairs, there does not seem to be a tendency for these internal relative price trends to
converge as we look at longer horizons. Finally, the conclusions are similar when looking
across the columns.
The exception for Hong Kong is hinted at in the MSE decompositions. In the first
panel in Figure 2 the curve begins near 100% and drops to about 80% before beginning to rise
12
toward 100% again. This is one of the most pronounced movements in all the figures.
However, since the curve remains near 100% overall, it must be the case that variation in
traded goods prices dwarfs this internal price difference. Overall then, Table 4 supports the
conclusion that cross-country offsetting movements in internal relative prices have been the
exception rather than the rule – at least for the horizons considered in this paper.
4. Conclusion
This study examined the proportion of real exchange rate movements that can be
attributed to movements in the relative price of non-traded goods among twenty-one bilateral
Asian-Pacific real exchange rates, for as far back historically as the data permits. Three
decompositions of real exchange rate movements were examined: drift, variance, and mean
squared error. Following Engel (1999), the decomposition is done at all possible horizons that
the data allow – from one month up to 25 years.
The evidence presented here is generally consistent with that from high-income U.S.
based real exchange rates. In particular, relative prices of non-traded goods appear to account
for virtually none of the movement of Pacific Rim real exchange rates. This pattern is
unaffected by the cross-sectional variation in either income level or the degree of openness
present among these Pacific-Rim economies.
The exceptions to these results are in accounting for movements in real exchange
rates. For a third of the bilateral Asian-Pacific real exchange rates studied, variation in the
relative price of non-traded goods account for less than 60% of variation in the real exchange
rate. This is very different from what Engel (1999) found in his sample of U.S. bilateral
exchange rates. The second exception to the previous evidence is for Hong Kong – especially
during the recent fixed exchange rate regime. In this case the importance of variation in
relative traded goods prices declines from seventy-five percent to fifty-percent (at six years)
before rising back to seventy-five percent. This is reminiscent of the result found by Mendoza
(2000) for the Mexican peso under managed exchange rate periods. However, in contrast to
his results, the other two pegged, or managed, currencies studied here do not demonstrate this
pattern. Moreover, the pattern exhibited by the Hong Kong dollar is more protracted (the
13
maximum impact of non-traded goods relative prices occurs at six years), and is smaller than
for the case of the peso. Additionally, statistically significant differences in trends in internal
relative price movements occur only in the case of Hong Kong. These differences suggest
that examining more cases of purely fixed exchange rates in future research may prove fruitful
in uncovering more cross-country variation in these accounting exercises.
Finally, one issue mentioned only briefly is that of measurement error. If traded goods
are bundled with non-traded services, and deviations in the former are stationary while those
in the latter are non-stationary, then the decompositions examined in this paper need to be
interpreted with caution. Indeed, general problems with aggregate price indexes have been
noted in two recent papers examining persistence. First, Imbs, Mumtaz, Ravn, and Rey (2002)
suggest that heterogeneity among the products within the aggregate price index seriously bias
persistence estimates upwards. In particular, OLS weights the more persistent component
more heavily, thus biasing the persistence estimate. They term this an aggregation bias. This
bias could, in principle, be a problem for the accounting exercises in this paper also. Second,
an additional bias is suggested by Taylor (2000). He argues that persistence estimates are
biased upward due to ‘temporal’ aggregation biases – the fact that all prices are not collected at
the same time during the month. Both of these biases suggest the data is poor – and it clearly
influences the inferences one makes.
Bibliography
Canzoneri, Matthew, R. Cumby, and B. Diba, 1999, “Relative Labor Productivity and the Real
Exchange Rate in the Long Run: Evidence for a Panel of OECD Countries”, Journal of
International Economics, 47, pp. 245-66.
Chinn, Menzie, 1996. “Asian Pacific Real Exchange Rates and Relative Prices”, University of
California, Santa Cruz working paper.
Chinn, Menzie, and Louis Johnston, 1999, “The Impact of Productivity Differentials on Real
Exchange Rates: Beyond the Balassa-Samuelson Framework”, University of California,
Santa Cruz working paper.
Cochrane, John, 1988, “How Big is the Unit Root Component of GNP?”, Journal of Political
Economy, 96, pp. 893-920.
Engel, Charles, 1999, “Accounting for U.S. Real Exchange Rate Changes”, Journal of
Political Economy, 107, pp. 507-38.
Frankel, Jeffrey and Andrew Rose, 2002, “An Estimate of the Effect of Common Currencies
on Trade and Income”, Quarterly Journal of Economics, 117 (2), pp. 437-466.
Froot, Kenneth, and K. Rogoff, 1995, “Perspectives on PPP and Long-Run Real Exchange
Rates”, in Handbook of International Economics, Volume 3, Gene Grossman and K.
Rogoff (eds.), Elsevier, New York.
Imbs, Jean, H. Mumtaz, M. Ravn, and H. Rey, 2002, “PPP Strikes Back: Aggregation and the
Real Exchange Rate”, NBER Working Paper # 9372.
Isard, Peter and S. Symansky, 1996, “Long-Run Movements in Real Exchange Rates”, in
Exchange Rate Movements and their Impact on Trade and Investment in the
APEC Region, pp. 7-28, Occasional Paper 145, International Monetary Fund.
Kakkar, Vikas, and M. Ogaki (1999), “Real Exchange Rates and Nontradables: A Relative Price
Approach”, Journal of Empirical Finance, 6, pp. 193-215.
Krugman, P. and M. Obstfeld, 1997, International Economics: Theory and Policy, 4th
edition. Addison Wesley Longman, Inc., New York.
Mendoza, Enrique G., 2000, “On the Instability of Variance Decompositions of the Real
Exchange Rate Across Exchange Rate Regimes: Evidence from Mexico and the United
States”, NBER Working Paper # 7768.
Obstfeld, M., and K. Rogoff, 2000, “The Six Major Puzzles in International Macroeconomics:
Is There a Common Cause?”, NBER Macroeconomics Annual, Vol. 15.
Parsley, David, and S. Wei, 2002, “Currency Arrangements and Goods Market Integration: A
Price Based Approach”, Vanderbilt University working Paper.
Rose, Andrew, “One Money, One Market: Estimating the Effect of Common Currencies on
Trade”, Economic Policy, 30, pp. 9-48.
Strauss, Jack, 1999, “Productivity Differentials, the Relative Price of Non-Tradables and Real
Exchange Rates, Journal of International Money and Finance, 18, pp. 383-409.
Taylor, Alan, 2001, “Potential Pitfalls for the Purchasing-Power Parity Puzzle? Sampling and
Specification Biases in Mean-Reversion Tests of the Law of One Price,” Econometrica,
69, pp. 473-98
Table 1: Summary Data
RGDP
Growth*
Real Per Capita GDP
Relative to the U.S.**
Exports + Imports
Percent of GDP
Hong Kong
6.7
90.6
286
Korea
6.2
42.1
61
Malaysia
3.7
31.0
156
Singapore
7.3
72.1
341
Taiwan
7.6
45.1
89
Thailand
6.1
21.6
77
Average
6.3
42.4
145
Canada
2.8
90.6
54
Denmark
2.3
80.7
66
Finland
2.3
67.3
53
France
2.2
78.5
45
Germany
2.7
87.0
60
Norway
3.3
73.6
79
Japan
2.8
85.8
18
Average
2.6
78.8
54
Engel’s Sample
Sources: International Financial Statistics (column 1), and Summers and Heston, Penn World Tables,
version 5.6 (columns 2 and 3).
* RGDP growth is computed as average annual growth, for the longest available period from 19751999.
** 1992 values are used except for Korea (1991), and Taiwan (1990).
Table 2: Fraction of Drift in Real Exchange Rate
Attributable to Traded Goods
(All Horizons)
U.S Dollar rates
Hong Kong dollar/US dollar
Korea won/US dollar
Malaysian ringgit/US dollar
Singapore dollar/US dollar
Taiwan dollar/US dollar
Thailand baht/US dollar
Observations
157
301
121
193
229
121
Traded Goods Component
0.512
0.906
0.882
0.556
0.990
0.906
Cross rates
Hong Kong dollar/Korean won
Hong Kong dollar/Malaysian ringgit
Hong Kong dollar/Singapore dollar
Hong Kong dollar/Taiwan dollar
Hong Kong dollar/Thailand baht
Korea won/Malaysian ringgit
Korea won/Singapore dollar
Korea won/Taiwan dollar
Korea won/Thailand baht
Malaysian ringgit/Singapore dollar
Malaysian ringgit/Taiwan dollar
Malaysian ringgit/Thailand baht
Singapore dollar/Taiwan dollar
Singapore dollar/Thailand baht
Taiwan/Thailand baht
229
121
193
229
121
121
193
229
121
121
121
121
193
121
121
0.568
0.844
0.999
0.941
0.150
0.485
0.836
0.801
0.907
0.227
0.726
0.849
0.866
0.581
0.904
Table 3: Average Drift Contribution to
MSE of Real Exchange Rates
U.S Dollar rates
Hong Kong dollar/US dollar
Korea won/US dollar
Malaysian ringgit/US dollar
Singapore dollar/US dollar
Taiwan dollar/US dollar
Thailand baht/US dollar
Drift(xt+yt)/MSE(xt+yt)
0.159
0.184
0.058
0.060
0.074
0.052
Cross rates
Hong Kong dollar/Korean won
Hong Kong dollar/Malaysian ringgit
Hong Kong dollar/Singapore dollar
Hong Kong dollar/Taiwan dollar
Hong Kong dollar/Thailand baht
Korea won/Malaysian ringgit
Korea won/Singapore dollar
Korea won/Taiwan dollar
Korea won/Thailand baht
Malaysian ringgit/Singapore dollar
Malaysian ringgit/Taiwan dollar
Malaysian ringgit/Thailand baht
Singapore dollar/Taiwan dollar
Singapore dollar/Thailand baht
Taiwan/Thailand baht
0.178
0.063
0.039
0.075
0.060
0.039
0.107
0.044
0.034
0.021
0.023
0.014
0.088
0.027
0.027
Table 4: Relative Movement of α ( p N − p T ) and β ( p N * − p T * )
(
)
(
)
Ho: ∆k α p N − p T - ∆k β p N * − p T * =0
U.S Dollar rates
Hong Kong dollar/US dollar
Korea won/US dollar
Malaysian ringgit/US dollar
Singapore dollar/US dollar
Taiwan dollar/US dollar
Thailand baht/US dollar
k=1
p-value Obs.
0.000
230
0.606
302
0.364
122
0.397
192
0.484
230
0.091
122
k=6
p-value Obs.
0.001 38
0.487 50
0.052 20
0.238 32
0.421 38
0.104 20
k=12
p-value Obs.
0.005
19
0.533
25
0.117
10
0.307
16
0.232
19
0.108
10
Cross rates
Hong Kong dollar/Korean won
Hong Kong dollar/Malaysian ringgit
Hong Kong dollar/Singapore dollar
Hong Kong dollar/Taiwan dollar
Hong Kong dollar/Thailand baht
Korea won/Malaysian ringgit
Korea won/Singapore dollar
Korea won/Taiwan dollar
Korea won/Thailand baht
Malaysian ringgit/Singapore dollar
Malaysian ringgit/Taiwan dollar
Malaysian ringgit/Thailand baht
Singapore dollar/Taiwan dollar
Singapore dollar/Thailand baht
Taiwan/Thailand baht
(
)
(
0.048
0.000
0.173
0.103
0.000
0.674
0.172
0.644
0.864
0.378
0.466
0.414
0.874
0.223
0.243
) (
230
122
192
230
122
122
192
230
122
120
122
122
192
120
122
)
0.014
0.000
0.061
0.011
0.000
0.601
0.084
0.572
0.857
0.322
0.330
0.456
0.817
0.175
0.051
38
20
32
38
20
20
32
38
20
20
20
20
32
20
20
0.039
0.002
0.076
0.026
0.006
0.519
0.124
0.529
0.856
0.420
0.079
0.417
0.779
0.222
0.064
19
10
16
19
10
10
16
19
10
10
10
10
16
10
10
Note: ∆k α p N − p T ≡ α ptN − ptT − α ptN−k − ptT−k . The table presents significance levels and
associated number of observations for values of k=1, 6, and 12.
Appendix 1: Consumer Price Data
Hong Kong
CPI Housing Rent: private
CPI Misc Services
CPI Durable goods
CPI Misc Goods
CPI All items
1981.01 – 2000.03
Korea
CPI Housing: House Rent
CPI Medical Care: Medical Services
CPI Food
less CPI Food: Eating Out
CPI All items
1975.01 – 2000.03
Malaysia
CPI Gross Rent, Fuel and Power
CPI Services
CPI Durable Goods
CPI Semi-durable Goods
CPI Non Durable Goods
CPI All items
1990.01 – 2000.03
Singapore
CPI Housing: Accommodation
CPI Misc: Household Services
CPI: Non Cooked food
CPI All items
1984.01 – 2000.01
Taiwan
CPI Housing
CPI Services
CPI Commodity Group including food
CPI All items
1981.01 – 2000.03
Thailand
CPI Housing and Furnishing: Shelter
CPI Health and Personal Care
CPI Food and Beverages
CPI All items
1990.01 – 2000.03
United States
CPI Housing
CPI All Commodities
1975.01 – 2000.03
All CPI data are taken from the CEIC database. Nominal exchange rates are expressed in
domestic currency per U.S. dollar and are period averages, also from CEIC.
Appendix 2: Average Covariance between x and y
Components of Real Exchange Rates
Cov(xt-xt-n,yt-yt-n)
U.S Dollar rates
Hong Kong dollar/US dollar
Korea won/US dollar
Malaysian ringgit/US dollar
Singapore dollar/US dollar
Taiwan dollar/US dollar
Thailand baht/US dollar
Average
Minimum
Maximum
0.00099
0.00026
-0.00054
0.00053
0.00047
0.00052
-0.00075
-0.00246
-0.00156
-0.00099
-0.00031
-0.00049
0.00286
0.00248
0.00006
0.00178
0.00180
0.00215
0.00285
-0.00213
-0.00100
0.00380
-0.00054
-0.00037
-0.00128
-0.00097
-0.00053
-0.00076
-0.00086
-0.00039
-0.00074
-0.00009
-0.00030
-0.00070
-0.00520
-0.00200
-0.00082
-0.00233
-0.00113
-0.00348
-0.00223
-0.00136
-0.00201
-0.00182
-0.00117
-0.00156
0.00083
0.00148
0.00802
0.00147
0.00000
0.00960
0.00190
0.00020
0.00147
0.00063
0.00000
0.00054
0.00006
0.00019
0.00000
0.00058
0.00071
Cross rates
Hong Kong dollar/Korean won
Hong Kong dollar/Malaysian ringgit
Hong Kong dollar/Singapore dollar
Hong Kong dollar/Taiwan dollar
Hong Kong dollar/Thailand baht
Korea won/Malaysian ringgit
Korea won/Singapore dollar
Korea won/Taiwan dollar
Korea won/Thailand baht
Malaysian ringgit/Singapore dollar
Malaysian ringgit/Taiwan dollar
Malaysian ringgit/Thailand baht
Singapore dollar/Taiwan dollar
Singapore dollar/Thailand baht
Taiwan dollar/Thailand baht
Figure 1: Variance Decompositions, domestic currency per U.S. dollar
(Percent of variance in real exchange rate accounted for by variance in relative price of traded goods)
HONG KONG DOLLAR
1.25
SINGAPORE DOLLAR
1.25
1.00
1.00
0.75
0.75
0.50
0.50
0.25
0.25
0.00
0.00
50
100
150
200
KOREAN WON
1.25
25
50
1.00
0.75
0.75
0.50
0.50
0.25
0.25
0.00
100
125
150
175
TAIWAN DOLLAR
1.25
1.00
75
0.00
50
100
150
200
250
MALAYSIAN RINGGIT
1.25
25
50
1.00
0.75
0.75
0.50
0.50
0.25
0.25
0.00
100
125
150
175
THAILAND BAHT
1.25
1.00
75
0.00
20
40
60
80
100
20
Horizon
40
60
80
100
200
Figure 1: Variance Decompositions, bilateral cross rates
(Percent of variance in real exchange rate accounted for by variance in relative price of traded goods)
1.25
HONG KONG DOLLAR PER KOREAN WON
1.25
KOREAN WON PER MALAYSIAN RINGITT
1.25
1.00
1.00
1.00
0.75
0.75
0.75
0.50
0.50
0.50
0.25
0.25
0.25
0.00
0.00
50
1.25
100
150
0.00
200
HONG KONG DOLLAR PER MALAYSIAN RINGITT
20
1.25
40
60
80
100
20
KOREAN DOLLAR PER SINGAPORE DOLLAR
1.25
1.00
1.00
1.00
0.75
0.75
0.75
0.50
0.50
0.50
0.25
0.25
0.25
0.00
0.00
20
1.25
40
60
80
100
1.25
50
75
100
125
150
KOREAN DOLLAR PER TAIWAN DOLLAR
20
1.25
1.00
1.00
0.75
0.75
0.75
0.50
0.50
0.50
0.25
0.25
0.25
0.00
25
1.25
50
75
100
125
150
175
HONG KONG DOLLAR PER TAIWAN DOLLAR
50
75
100
125
150
175
200
KOREAN DOLLAR PER THAILAND BAHT
1.25
1.00
1.00
0.75
0.75
0.75
0.50
0.50
0.50
0.25
0.25
0.25
0.00
0.00
1.25
50
75
100
125
150
175
HONK KONG DOLLAR PER THAILAND BAHT
200
40
60
80
100
MALAYSIAN RINGITT PER SINGAPORE DOLLAR
1.25
1.00
1.00
0.75
0.75
0.75
0.50
0.50
0.50
0.25
0.25
0.25
0.00
0.00
40
60
80
100
60
80
100
50
75
100
125
150
175
SINGAPORE DOLLAR PER THAILAND BAHT
20
1.00
20
40
0.00
20
1.25
100
SINGAPORE DOLLAR PER TAIWAN DOLLAR
25
1.00
25
80
0.00
25
1.25
60
MALAYSIAN RINGITT PER THAILAND BAHT
175
1.00
0.00
40
0.00
25
HONG KONG DOLLAR PER SINGAPORE DOLLAR
MALAYSIAN RINGITT PER TAIWAN DOLLAR
40
60
80
100
TAIWAN DOLLAR PER THAILAND BAHT
0.00
20
40
60
Horizon
80
100
20
40
60
80
100
Figure 2: Mean Squared Error Decompositions, domestic currency per U.S. dollar
(Percent of total variation (MSE) in real exchange rate accounted for by MSE of relative price of traded goods)
HONG KONG DOLLAR
1.25
SINGAPORE DOLLAR
1.25
1.00
1.00
0.75
0.75
0.50
0.50
0.25
0.25
0.00
0.00
50
100
150
200
KOREAN WON
1.25
25
50
1.00
0.75
0.75
0.50
0.50
0.25
0.25
0.00
100
125
150
175
TAIWAN DOLLAR
1.25
1.00
75
0.00
50
100
150
200
250
MALAYSIAN RINGGIT
1.25
50
1.00
0.75
0.75
0.50
0.50
0.25
0.25
0.00
150
200
THAILAND BAHT
1.25
1.00
100
0.00
20
40
60
80
100
20
Horizon
40
60
80
100
Figure 2: Mean Squared Error Decompositions, bilateral cross rates
(Percent of total variation (MSE) in real exchange rate accounted for by MSE of relative price of traded goods)
1.25
HONG KONG DOLLAR PER KOREAN WON
1.25
KOREAN WON PER MALAYSIAN RINGITT
1.25
1.00
1.00
1.00
0.75
0.75
0.75
0.50
0.50
0.50
0.25
0.25
0.25
0.00
0.00
50
1.25
100
150
0.00
200
20
HONG KONG DOLLAR PER MALAYSIAN RINGITT
1.25
40
60
80
100
20
KOREAN DOLLAR PER SINGAPORE DOLLAR
1.25
1.00
1.00
1.00
0.75
0.75
0.75
0.50
0.50
0.50
0.25
0.25
0.25
0.00
0.00
20
1.25
40
60
80
100
1.25
50
75
100
125
150
175
KOREAN DOLLAR PER TAIWAN DOLLAR
20
1.25
1.00
1.00
1.00
0.75
0.75
0.75
0.50
0.50
0.50
0.25
0.25
0.25
0.00
0.00
25
1.25
50
75
100
125
150
50
1.25
100
150
200
KOREAN DOLLAR PER THAILAND BAHT
1.25
1.00
1.00
0.75
0.75
0.75
0.50
0.50
0.50
0.25
0.25
0.25
0.00
0.00
1.25
100
150
200
HONK KONG DOLLAR PER THAILAND BAHT
40
60
80
100
MALAYSIAN RINGITT PER SINGAPORE DOLLAR
1.25
1.00
1.00
0.75
0.75
0.75
0.50
0.50
0.50
0.25
0.25
0.25
0.00
0.00
40
60
80
100
40
50
60
80
100
75
100
125
150
175
SINGAPORE DOLLAR PER THAILAND BAHT
20
1.00
20
100
0.00
20
1.25
80
SINGAPORE DOLLAR PER TAIWAN DOLLAR
25
1.00
50
60
0.00
175
HONG KONG DOLLAR PER TAIWAN DOLLAR
40
MALAYSIAN RINGITT PER THAILAND BAHT
0.00
25
HONG KONG DOLLAR PER SINGAPORE DOLLAR
MALAYSIAN RINGITT PER TAIWAN DOLLAR
20
40
Horizon
60
80
100
0.00
40
60
80
100
TAIWAN DOLLAR PER THAILAND BAHT
20
40
60
80
100
Figure 3: Variance and MSE Decompositions
(Fixed and semi-Fixed Exchange Rate Sample Only)
1.25
Variance
Decompositions
HONG KONG DOLLAR
MALAYSIAN RINGGIT
1.25
1.00
1.00
1.00
0.75
0.75
0.75
0.50
0.50
0.50
0.25
0.25
0.25
0.00
0.00
50
1.25
MSE
Decompositions
1.25
100
150
HONG KONG DOLLAR
0.00
20
1.25
40
60
80
100
MALAYSIAN RINGGIT
20
1.25
1.00
1.00
1.00
0.75
0.75
0.75
0.50
0.50
0.50
0.25
0.25
0.25
0.00
0.00
50
100
150
THAILAND BAHT
40
60
80
100
THAILAND BAHT
0.00
20
40
60
Horizon
80
100
20
40
60
80
100
Appendix Figure 1: Nominal Exchange Rates, domestic currency per U.S. dollar (logs)
HONG KONG DOLLAR PER US DOLLAR
2.1
SINGAPORE DOLLAR PER US DOLLAR
0.9
0.8
2.0
0.7
1.9
0.6
1.8
0.5
1.7
0.4
1.6
0.3
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
KOREAN WON PER US DOLLAR
7.50
1984
1986
3.7
7.00
3.6
6.75
3.5
6.50
3.4
6.25
3.3
6.00
1990
1992
1994
1996
1998
TAIWAN DOLLAR PER US DOLLAR
3.8
7.25
1988
3.2
1975
1978
1981
1984
1987
1990
1993
1996
MALAYSIAN RINGGIT PER US DOLLAR
1.52
1999
1982
1984
1986
1988
1990
1992
1994
1996
THAILAND BAHT PER US DOLLAR
4.0
3.9
1.44
3.8
1.36
3.7
1.28
3.6
1.20
3.5
1.12
3.4
1.04
3.3
0.96
3.2
0.88
3.1
1990
1992
1994
1996
1998
1990
1992
1994
1996
1998
1998
Appendix Figure 1: Nominal Exchange Rates, bilateral cross rates (logs)
HONG KONG DOLLAR PER KOREAN WON
-4.4
KOREAN WON PER MALAYSIAN RINGITT
6.12
MALAYSIAN RINGITT PER TAIWAN DOLLAR
-2.00
-2.05
-4.6
6.00
-4.8
5.88
-5.0
5.76
-5.2
5.64
-2.10
-2.15
-2.20
-2.25
-2.30
-2.35
-2.40
-5.4
5.52
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
HONG KONG DOLLAR PER MALAYSIAN RINGITT
1.2
-2.45
1990
1992
1993
1994
1995
1996
1997
1998
1999
KOREAN DOLLAR PER SINGAPORE DOLLAR
7.0
1.1
1991
1991
1992
1993
1994
1995
1996
1997
1998
1999
1998
1999
MALAYSIAN RINGITT PER THAILAND BAHT
-2.20
-2.25
6.8
1.0
1990
-2.30
6.6
0.9
-2.35
6.4
0.8
-2.40
6.2
0.7
-2.45
6.0
0.6
0.5
-2.50
5.8
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
HONG KONG DOLLAR PER SINGAPORE DOLLAR
1.8
-2.55
1984
1986
1988
1990
1992
1994
1996
1998
KOREAN DOLLAR PER TAIWAN DOLLAR
4.00
1992
1993
1994
1995
1996
1997
-2.64
3.68
1.6
1991
SINGAPORE DOLLAR PER TAIWAN DOLLAR
-2.56
3.84
1.7
1990
-2.72
3.52
1.5
-2.80
3.36
1.4
-2.88
3.20
1.3
3.04
1.2
2.88
1984
1986
1988
1990
1992
1994
1996
1998
HONG KONG DOLLAR PER TAIWAN DOLLAR
-1.1
-2.96
-3.04
1982
1984
1986
1988
1990
1992
1994
1996
1998
KOREAN DOLLAR PER THAILAND BAHT
3.60
1984
-2.6
-1.2
3.55
-2.7
-1.3
3.50
-2.8
-1.4
3.45
-1.5
3.40
-1.6
3.35
-1.7
3.30
-1.8
3.25
-1.9
1984
1986
1988
1990
1992
1994
1996
1998
HONK KONG DOLLAR PER THAILAND BAHT
-1.1
-1.3
0.80
-1.4
-1.8
0.48
-1.9
0.40
-2.0
0.32
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
1998
-3.4
-3.5
1991
1992
1993
1994
1995
1996
1997
1998
1999
MALAYSIAN RINGITT PER SINGAPORE DOLLAR
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
TAIWAN DOLLAR PER THAILAND BAHT
0.1
-0.0
-0.1
-0.2
0.56
-1.7
1996
-3.3
0.64
-1.6
1994
-3.2
0.72
-1.5
1992
-3.1
0.96
0.88
1990
-3.0
1990
-1.2
1988
-2.9
3.20
1982
1986
SINGAPORE DOLLAR PER THAILAND BAHT
-0.3
-0.4
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
-0.5
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
Appendix Figure 2: Real Exchange Rates, domestic currency per U.S. dollar (logs)
HONG KONG DOLLAR PER US DOLLAR
2.5
SINGAPORE DOLLAR PER US DOLLAR
0.70
0.65
2.4
0.60
2.3
0.55
2.2
0.50
0.45
2.1
0.40
2.0
0.35
1.9
0.30
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
KOREAN WON PER US DOLLAR
7.3
1984
1986
1990
1992
1994
1996
1998
TAIWAN DOLLAR PER US DOLLAR
3.7
7.2
1988
3.6
7.1
3.5
7.0
6.9
3.4
6.8
3.3
6.7
6.6
3.2
1975
1978
1981
1984
1987
1990
1993
1996
MALAYSIAN RINGGIT PER US DOLLAR
1.5
1999
1982
1984
1986
1988
1990
1992
1994
1996
THAILAND BAHT PER US DOLLAR
3.9
3.8
1.4
3.7
1.3
3.6
1.2
3.5
1.1
3.4
1.0
3.3
0.9
3.2
0.8
3.1
1990
1992
1994
1996
1998
1990
1992
1994
1996
1998
1998
Appendix Figure 2: Real Exchange Rates, bilateral cross rates (logs)
HONG KONG DOLLAR PER KOREAN WON
-4.5
KOREAN WON PER MALAYSIAN RINGITT
6.06
6.00
-2.10
-4.7
5.94
-2.15
5.88
-2.20
5.82
-2.25
5.76
-2.30
-5.2
5.70
-2.35
-5.3
5.64
-2.40
-5.4
5.58
-4.8
-4.9
-5.0
-5.1
1981
1.2
1984
1987
1990
1993
1996
1999
HONG KONG DOLLAR PER MALAYSIAN RINGITT
1.0
6.6
0.9
6.5
0.8
6.4
0.7
6.3
0.6
6.2
0.5
6.1
0.4
6.0
1.9
1992
1994
1996
1998
HONG KONG DOLLAR PER SINGAPORE DOLLAR
1.7
3.7
1.6
3.6
1.5
3.5
1.4
3.4
1.3
3.3
1984
-0.99
1986
1988
1990
1992
1994
1996
1998
HONG KONG DOLLAR PER TAIWAN DOLLAR
1998
1990
-2.10
-2.45
1986
1988
1990
1992
1994
1996
1998
KOREAN DOLLAR PER TAIWAN DOLLAR
3.45
3.30
-1.62
3.25
1982
1984
1986
1988
1990
1992
1994
1996
HONK KONG DOLLAR PER THAILAND BAHT
-1.20
1998
1998
SINGAPORE DOLLAR PER TAIWAN DOLLAR
-3.0
-3.1
1984
1986
1988
1990
1992
1994
1996
1998
KOREAN DOLLAR PER THAILAND BAHT
1984
-2.7
1986
1988
1990
1992
1994
1996
1998
SINGAPORE DOLLAR PER THAILAND BAHT
-2.8
-2.9
-3.0
-3.1
-3.2
-3.3
1992
1994
1996
1998
MALAYSIAN RINGITT PER SINGAPORE DOLLAR
0.9
-1.32
1996
-2.9
1990
1.0
1994
-2.8
3.35
-1.53
1992
-2.7
3.40
-1.44
1990
-2.6
3.55
3.50
1998
-2.40
3.60
-1.35
1996
-2.35
1982
-1.26
1994
-2.30
3.65
-1.17
1992
MALAYSIAN RINGITT PER THAILAND BAHT
-2.25
3.70
-1.08
-1.08
1996
-2.20
3.9
3.8
1994
-2.15
1984
1.8
1992
KOREAN DOLLAR PER SINGAPORE DOLLAR
6.8
6.7
1990
-2.45
1990
1.1
MALAYSIAN RINGITT PER TAIWAN DOLLAR
-2.05
-4.6
1990
1992
1994
1996
1998
TAIWAN DOLLAR PER THAILAND BAHT
0.2
0.1
0.8
-0.0
-1.44
0.7
-1.56
-0.1
0.6
-1.68
-1.80
0.5
-1.92
0.4
1990
1992
1994
1996
1998
-0.2
-0.3
1990
1992
1994
1996
1998
1990
1992
1994
1996
1998
Appendix Figure 3: Internal Relative Prices (P N P T )
HONG KONG & US
0.6
HK
US
0.5
SINGAPORE & US
0.4
SI
US
0.3
0.4
0.3
0.2
0.2
0.1
0.1
-0.0
-0.1
0.0
-0.2
-0.3
-0.1
1981
1984
1987
1990
1993
1996
1999
1984
KOREA & US
0.25
KO
US
1986
1988
0.30
0.15
0.25
1992
1994
1996
1998
TAIWAN & US
0.35
0.20
1990
TA
US
0.20
0.10
0.15
0.05
0.10
0.00
0.05
-0.05
0.00
-0.10
-0.05
-0.15
-0.10
1975
1978
1981
1984
1987
1990
1993
1996
1999
MALAYSIA & US
0.050
MA
US
0.025
1981
1984
1987
1990
1993
1996
THAILAND & US
0.150
TH
US
0.125
0.100
0.075
0.000
0.050
0.025
-0.025
0.000
-0.025
-0.050
-0.050
-0.075
-0.075
1990
1992
1994
1996
1998
1999
1990
1992
1994
1996
1998
Appendix Figure 3: Internal Relative Prices (P N P T )
HONG KONG & KOREA
0.6
HK
KO
0.5
0.4
KOREA & MALAYSIA
0.04
KO
MA
0.02
-0.02
0.2
MA
TA
0.10
0.00
0.3
MALAYSIA & TAIWAN
0.15
0.05
-0.04
0.1
0.00
-0.06
-0.0
-0.08
-0.1
-0.05
-0.10
-0.2
-0.3
-0.12
1981
1984
1987
1990
1993
1996
1999
HONG KONG & MALAYSIA
0.36
-0.10
1990
HK
MA
1992
1994
1996
1998
1990
KOREA & SINGAPORE
0.4
KO
SI
1992
1994
1996
1998
MALAYSIA & THAILAND
0.15
MA
TH
0.3
0.24
0.10
0.2
0.12
0.1
0.00
0.05
0.0
0.00
-0.12
-0.1
-0.24
-0.2
1990
1992
1994
1996
1998
HONG KONG & SINGAPORE
0.5
-0.05
1984
HK
SI
1988
0.30
0.3
0.24
0.2
0.18
0.1
0.12
-0.0
0.06
-0.1
0.00
-0.2
-0.06
1990
1992
1994
1996
1998
1990
KOREA & TAIWAN
0.36
0.4
-0.3
1986
KO
TA
1986
1988
1990
1992
1994
1996
1998
0.6
HK
TA
0.5
0.3
-0.0
1987
1990
1993
1996
1984
1999
KOREA & THAILAND
0.15
KO
TH
-0.3
1984
1987
1990
1993
1996
1999
0.36
HK
TH
0.24
1998
0.150
SI
TH
0.125
-0.050
-0.075
1992
1994
1996
1998
1990
MALAYSIA & SINGAPORE
0.12
MA
SI
0.10
1992
1994
1996
1998
TAIWAN & THAILAND
0.15
TA
TH
0.10
0.05
0.04
0.02
0.00
0.00
-0.02
-0.12
1996
0.000
0.06
0.00
1994
-0.025
0.08
0.12
1992
0.025
1990
HONK KONG & THAILAND
1990
SINGAPORE & THAILAND
0.050
-0.15
1981
1988
0.075
-0.10
-0.2
1986
0.100
-0.05
-0.1
SI
TA
-0.1
1984
0.00
0.1
0.4
0.0
0.05
0.2
1998
0.1
0.10
0.4
1996
0.2
1981
HONG KONG & TAIWAN
1994
SINGAPORE & TAIWAN
0.3
-0.12
1984
1992
-0.05
-0.04
-0.24
1990
1992
1994
1996
1998
-0.06
1990
1992
1994
1996
1998
-0.10
1990
1992
1994
1996
1998