You are on page 1of 17

INTERNATIONAL ECONOMIC JOURNAL

Volume 11, Number 1, Spring 1997

51

ARE ASIAN MARKETS INTEGRATED? EVIDENCE FOR SIX


COUNTRIES VIS-A-VIS JAPAN
IMAD A. MOOSA*
La Trobe University
RAZZAQUE H. BHATTI
University of Azad Jammu and Kashmir

This paper presents some empirical evidence on the degree of integration between
the goods and financial markets of Japan and six Asian countries. The evidence is
obtained by testing two international parity conditions using unconventional
specifications: uncovered interest parity (UIP) and ex ante purchasing power parity
(PPP). The results of cointegration tests are strongly supportive of the two conditions
over the period 1980-1994. The paper concludes that Asian goods and financial
markets have reached a high level of integration. [G14]

1. INTRODUCTION
Over the last fifteen years, most Asian countries are thought to have achieved a
high degree of integration among their markets for goods, capital and foreign
exchange. This is because most of these countries deregulated and liberalised their
domestic markets during the 1980s. Financial liberalisation has been undertaken for
the purpose of increasing the efficiency of the domestic financial system by
liberalising interest rates, reducing controls on credit, enhancing competition among
financial institutions and encouraging the creation and development of money and
capital markets. For example, Singapore largely liberalised its financial sector and
abolished capital controls in the mid 1970s, while Hong Kong, Japan, Malaysia and
the Philippines embarked on the liberalisation scheme in the early 1980s. On the
other hand, Korea undertook more gradual measures towards financial liberalisation
which were intensified in the second half of the 1980s. Together with this move
towards the liberalisation of domestic financial markets, most Asian countries also
undertook measures to relax international capital controls and to adopt more flexible
exchange rate arrangements. For example, Japan moved to the flexible exchange rate
system in 1973, while Malaysia and Singapore began to experiment with controlled
floating in the early 1970s. Furthermore, most of the other Asian countriessuch as
Korea, the Philippines, Singapore and Taiwanmaintained fixed exchange rate
*We are grateful to three anonymous referees for their comments on a previous version of
the paper. We would also like to thank Lee Smith for her inquisitive reading of the paper and

52

I. A. MOOSA AND R. H. BHATTI

arrangements by pegging their currencies to the U.S. dollar following its float in 1971.
However, these arrangements were abolished when most of these countries delinked
their currencies from the U.S. dollar in preference for controlled floating of the
effective exchange rate (Korea in 1980, the Philippines in 1984 and Taiwan in 1979).
Moreover, by the early 1980s, Singapore eliminated virtually all restrictions on capital
flows while Malaysia abolished almost all controls on international capital flows that
were not financed by local borrowing.
The extent of international market integration greatly affects the behaviour of
exchange and interest rates across countries, which in turn have crucial implications
for the degree to which the domestic monetary authorities can pursue independent
monetary policies. There is little dispute over the proposition that the more integrated
the international markets for goods, capital and foreign exchange the more limited is
the scope for pursuing independent domestic monetary policies. For example, if
goods and capital move around to eliminate the differential between prices and
interest rates across countries, then the domestic monetary authorities will have no
control over their real exchange and interest rates relative to those of other countries,
limiting the impact of their stabilisation policies. Therefore, it is necessary for policy
makers to take full account of the possible repercussions of international market
integration.
The objective of this paper is to examine integration between the Japanese
markets and those of six Asian countriesHong Kong, Korea, Malaysia, the
Philippines, Singapore and Taiwanover the period 1980-1994. The remainder of
the paper is organised as follows. Section 2 gives a brief description of the
hypotheses used to test the extent of market integration, while section 3 presents the
specification of the models to be tested. Section 4 deals with the data, methodology
and the results of empirical testing, and we end up with some concluding remarks in
the final section.
2. THE HYPOTHESES
Economists have employed several procedures to test (capital) market integration
and capital mobility, two terms that are often used interchangeably.1 The degree of
international market integration can be tested directly by comparing nominal and real
returns on comparable goods and financial assets across countries. Direct testing to
determine the degree of international market integration can be carried out by
examining the validity of various international parity conditions: purchasing power
parity (PPP), covered interest parity (CIP), uncovered interest parity (UIP) and real
The two terms are often used interchangeably because it is plausible to postulate that
capital moves more freely if capital markets are integrated. However, a careful reading of the
literature, as manifested by Goldstein et al. (1991) and Frankel (1993), reveals that while
capital mobility is not a necessary condition for market integration, the latter is conducive to the
former, while the former is a sufficient condition for the latter.
1

ARE ASIAN MARKETS INTEGRATED?

53

interest parity (RIP). While the PPP condition is based on a comparison of the returns
on identical goods, the other conditions are concerned with the returns on perfectly
substitutable financial assets across countries. Moreover, while CIP and UIP pertain
to nominal returns on financial assets denominated in different currencies, RIP
pertains to real returns. It must be noted that while PPP is a measure of integration
between goods markets, CIP and UIP indicate the degree of integration between
financial markets, and RIP is a joint measure of goods and financial market
integration.2
The extent of market integration can also be tested indirectly by examining the
degree of correlation between national saving and investment. The indirect test of
international market integration is found in the work of Feldstein and Horioka (1980)
and Feldstein (1983) who argue that in an internationally integrated financial market,
potentially infinite capital flows eliminate differentials among nominal and real rates
of return on identical assets, implying that a shortfall of saving in one country is
unlikely to restrict its volume of investment therein. This is because perfect mobility
of capital breaks the link between national saving and national investment and,
therefore, a fall in private saving or a deficit in the current account in a country is
unlikely to crowd out investment by raising the real cost of borrowing: rather, this
country can then borrow sufficient funds at the going world interest rate to make up
the difference. However, Frankel (1992) argues that there are several problems with
the saving-investment criterion: (1) the presence of cyclical movements may result in
strong correlation between national saving and investment; (2) national saving may
become endogenous if governments respond incipient current account imbalances
with policies to change public (or private) saving in such a way as to reduce the
imbalances; and (3) the correlation between saving and investment is reduced when
large countries are excluded from the sample, implying that the world real interest
rate will not be exogenous if the domestic country is large enough in world financial
markets (see also Tobin, 1983 and Murphy, 1984). The Feldstein-Horioka findings
have also been challenged by Bodman (1995) who re-examined the puzzle using
cointegration analysis and produced strong time series evidence for capital
m o b i l i t y .3
Aggarwal and Mougoue (1993) have proposed a test for a limited definition of
financial integration which they applied to Asian countries vis-a-vis Japan. The test is
based on a finding of cointegration, or otherwise, between the currencies of Japan, on
one hand, and Hong Kong, Malaysia, the Philippines and Singapore on the other.
2
Several testing techniques and model specifications have been used in conjunction with
these criteria. For example, Modjtahedi (1987) tested RIP using linear dynamic stochastic
representations, Boothe (1991) combined UIP with the term structure of interest rates, Moosa
and Bhatti (1995) tested RIP in a time varying parametric framework, while Moosa (1995)
tested both UIP and ex ante PPP using recursive and rolling regressions.
3
Bodman (1995:38) distinguishes between studies of capital mobility based on the
Feldstein-Horioka methodology and those based on rates of return by distinguishing between
real and financial capital.

54

I. A. MOOSA AND R. H. BHATTI

This test, however, is not a test for financial market integration in the broad sense but
rather a test for the existence of a yen bloc.4
This paper follows the direct approach to testing the degree of international market
integration. The degree of integration between goods markets is determined by
examining ex ante PPP, while integration between financial markets is determined by
examining UIP. The reason why this approach is adopted is that we are interested in
financial rather than real capital which is implied by the Feldstein-Horioka analysis.
Moreover, although RIP can be used as a test for the simultaneous integration of
capital and goods markets, the empirical failure of this condition cannot be readily
interpreted as implying the lack of integration of capital markets, goods markets or
both. Thus, more insight is gained by testing the two constituent components of RIP
(UIP and ex ante PPP) which indicate capital and goods market integration
respectively.5
3. MODEL SPECIFICATION
The UIP and ex ante PPP conditions are modelled along the lines proposed by
Bhatti and Moosa (1994, 1995) using specifications in levels rather than the
conventional first differences. This specification is conducive to employing
cointegration analysis while avoiding the spurious regression problem.6 Since these
specifications are derived in Moosa and Bhatti (1994, 1995) the treatment here will be
rather brief.
The UIP hypothesis postulates that in the presence of perfect capital mobility with
no capital controls, transaction costs and risk premia, the expected rate of change of
the spot exchange rate will be exactly equal to the nominal interest rate differential on
perfectly comparable financial assets denominated in different currencies across
countries. This condition is given by
(1 + )i = (1

+S e )(1 + )i

(1)

One important implication of a finding of cointegration between these currencies is that the
foreign exchange markets in these countries are cross-sectionally inefficient. This is because
according to Grangers Representation Theorem, cointegration implies and is implied by a valid
error correction representation. Since the latter includes a lagged error correction term, this
implies the possibility of predicting changes in one currency on the basis of changes in another
currency, an indication of cross-sectional inefficiency (see Moosa and A1-Loughani, 1994).
5
See Moosa (1995).
6
See, for example, Bodman (1995:46). Differencing, which renders the underlying
variables stationary and makes it possible to use conventional test statistics, became a
widespread econometric procedure following the rise of the popularity of the Box-Jenkins
methodology. The problem with this procedure, however, is that differencing causes a loss of
information, a problem that can be avoided by applying cointegration analysis to level data.
4

ARE ASIAN MARKETS INTEGRATED?

55

where i (i*) is the domestic (foreign) interest rate and


is the expected rate of change
of the spot exchange rate (where the spot exchange rate, S, is defined as the domestic
currency price of a unit of the foreign currency). The alternative specification is
derived by solving equation (1) for the expected spot exchange rate, Se (rather than the
expected rate of change of the spot exchange rate,
) to obtain

Se = F

(2)

where = S[(1 + i) / (1 + i*)] is the interest parity forward rate which is equal to the
actual forward exchange rate, F, if and only if the CIP condition holds precisely.7
Let us now re-rewrite equation (2) as
ste + 1 = ft

(3)

where ste + 1is the logarithm of the expected spot rate and
ft is the logarithm of the interest
parity forward rate. By allowing for the behaviour of the risk premium and
incorporating the rational expectations hypothesis the model can be written in a
testable form as8
st+1 = 0 + 1 ft + t+1

(4)

where t+1 is an error term reflecting the impact of news, and 0 is a constant term
reflecting the value of the risk premium as well as other factors such as transaction
costs. The UIP condition holds in a strong form if the restrictions 0 = 0 and 1 = 1
are not rejected.
The ex ante PPP hypothesis, which has been developed by Roll (1979) and Adler
and Lehman (1983), postulates that the expected rate of change in the nominal
exchange rate tends to be equal to the expected differential in inflation rates across
countries over the same holding period. Based on this hypothesis, Bhatti and Moosa
(1994) proposed a view of the exchange rate determination process which postulates
that if the nominal exchange rate moves to equalise the expected domestic and
foreign inflation rates, then the equilibrium nominal exchange rate is determined not
7
Equation (2) represents UIP in a deterministic form because it can be obtained by
combining CIP (F = ) with the unbiasedness hypothesis (F = Se).
8
The rational expectations hypothesis is incorporated by specifying the actual value of the
exchange rate to be equal to the expected value plus a random error.

56

I. A. MOOSA AND R. H. BHATTI

only by current relative prices, but also by the expected real exchange rate. The ex
ante PPP relationship, which is based on intertemporal commodity arbitrage, is
given by
e
e
st+1
= pt+1
- pe
t+1

(5)

where st+1 is the expected change in the logarithm of the spot exchange rate and
e
e
pt+1 (pt+1) is the expected change in the logarithm of the domestic (foreign) price
9
level. By incorporating the rational expectations hypothesis we obtain
st+1 - pt+1 + pt+1 =u t+1

(6)

where ut+1 is a composite expectations error term. By writing equation (6) in levels
and rearranging we obtain

st = pt - pt +s t+1 +p t+1
- pt+1

(7)

which can be written in a testable stochastic form as


st = + (p - p )t + qt+1 + vt

(8)

where qt+1 is the ex post real exchange rate (defined as qt+1 = st+1 + p*t+1 Pt+1)
realised at time t + 1 and vt is an error term that partially reflects expectations errors.
The hypothesis that the nominal exchange rate is determined not only by the current
price ratio but also by the expected real exchange rate will hold if the coefficient on
the expected real exchange rate, qt+1, is statistically different from zero. The other
relevant coefficient restrictions are = 0, = 1 and = 1.
An important issue here is that the specification of equations (4) and (8) embody
the hypotheses of risk neutrality and rational expectations. Economists have normally
attributed the failure of market efficiency to risk aversion (i.e. the existence of risk
premium) rather than to the failure of rational expectations. This tendency can be
attributed to two reasons: (i) it is difficult to conceive that market agents are irrational
(in the spirit of neoclassical economics), and (ii) testing the rational expectations
e
Hence pt+1
(p e
t+1) is approximately the expected domestic (foreign) inflation rate.

ARE ASIAN MARKETS INTEGRATED?

57

hypothesis is not an easy task because it requires the availability of survey data. The
failure of market efficiency normally leads economists to model the risk premium
either explicitly or as a function of the variance of exchange rate movements (Taylor,
1995). It is worth noting in this respect that the rational expectations hypothesis is not
necessary for testing the models represented by equations (4) and (8). If expectations
are formed in such a way that the error is stationary (as opposed to white noise as
required by the rational expectations hypothesis), then cointegration in models
involving expectational variables will imply cointegration in the models involving the
corresponding observed variables (equations 4 and 8). This is one more advantage of
using cointegration analysis, and hence of specifying the models in levels.
4. DATA, METHODOLOGY AND EMPIRICAL RESULTS
Testing the hypothesis of international market integration is carried out on the
basis of the UIP and ex ante PPP relationships implied by equations (4) and (8)
respectively. The sample data consists of quarterly observations on exchange rates,
consumer prices and 3-month treasury bill (or 3-month deposit) rates of Japan and six
other Asian countries - Hong Kong, Korea, Malaysia, the Philippines, Singapore and
Taiwan. The data sample, which covers the period extending between 1980:1 and
1994:4, was obtained from the IMFs International Financial Statistics as reported by
Datastream. The short-term interest rates used to test UIP are 3-month deposit rates
for Hong Kong and Malaysia and 3-month treasury bill rates for Taiwan, the
Philippines and Korea. To make the rates comparable, we have used the 3-month
deposit rate and the 3-month Gensaki rate for Japan.
Cointegration analysis is a technique that seems to be tailor-made for testing these
parity conditions as long-run relationships while allowing for short-run deviations
from equilibrium.10 In this case equations (4) and (8) can be viewed as attractors
towards which the points representing combinations of the underlying variables tend
to return to following a disturbance. The mechanism which moves the points towards
the attractor, thus re-establishing and maintaining equilibrium, is financial arbitrage in
the case of UIP and commodity arbitrage in the case of ex ante PPP. We may sum up
the advantages of using cointegration analysis in this study as follows: (i) it enables us
to avoid the loss of information resulting from differencing while distinguishing
between spurious and genuine long-run relationships, (ii) it produces superconsistent
estimates of the regression coefficients, and (iii) the stringent requirement of white
noise expectations errors can be replaced by the less stringent requirement of
stationary errors.
Before testing for cointegration, unit root testing is carried out to determine the
order of integration of the variables under investigation. For this purpose, the DickeyFuller (1979) test is used. As applied to the time series, yt, the statistic is the t
10
One of the earliest applications of cointegration anlysis was testing PPP (e.g., Taylor,
1988).

58

I. A. MOOSA AND R. H. BHATTI

ratio of in the regression equation


m

yt = + yt-1 +

y
i

+ vt

t-i

(9)

i=1

Since the augmentation terms are added to remove serial correlation, the criterion
that is adopted to determine the order of the augmented Dickey-Fuller test (the
value of m) is that it should be the smallest number necessary to remove serial
correlation.11
Testing for cointegration is carried out by applying the Phillips-Ouliaris
(1990)
tests to the residuals of the cointegrating regressions (4) and (8).
These tests, which were originally proposed by Phillips (1987), are more robust to a
wide variety of serial correlation, time dependent heteroscedasticity and regime
changes. The calculation of these statistics is based on the auxilliary regression

v t = v t-1 + tu

(10)

is the estimated value of in (10) then the

If

statistic is given by

Z = N( - )1 - (1 / 2)(sT2 - V2s)(N -2

2 -1
t-1 )

(11)

where N is the sample size. sV2 and sT2 are given by


N

sV2 =

-1

2
t

(12)

and
N

sT2 = N-1

ut2 +

2 -1N

w sk

s=1

ut ut-s
t=s+1

(13)

11
Too many augmentation terms will make equation (9) over-parameterised and reduce the
power of the test.

ARE ASIAN MARKETS INTEGRATED?

59

for some window such as wsk = 1 - s/(k + 1). The

Zt =

1/2

2
v t-1

( - )/s
1 T - (1/2 )(

sT2

2
Vs) sT

statistic is given by

-2
N

v 2t-1

12/

-1

(14)

Phillips and Ouliaris (1990) have demonstrated that these tests have limiting
distributions which are free of nuisance parameters and which, in the general case, are
the same as those of the Dickey-Fuller and t tests in the highly restrictive case of iid
errors.12 The critical values are tabulated in Phillips and Ouliaris (1990).13
A problem that will arise if the underlying variables are nonstationary is that
although the OLS estimates of the coefficients are superconsistent if the variables are
cointegrated, they are not fully efficient and their standard errors will not be
asymptotically normal. Therefore, the hypotheses implied by the coefficient
restrictions cannot be tested on the basis of the conventional t statistics. This problem
can be solved by correcting the conventional standard errors along the lines proposed
by West (1988). With respect to equation (8) this correction requires dividing the
conventional t statistics by s / (0) where

(0)=

1
N

vt2

i=1
m

(15)

and
m

[1

s = (0) + 2

j=1

-| j|/(m

v v

+ )]1

t t+| j |

(16)

t+| j |+1

An alternative procedure would be to introduce some dynamics into the equations to


increase the efficiency, in which case the conventional t statistics will be valid (see
Lim and Martin, 1995 for an excellent survey of these methods).14
When the estimates of sT2 are based on the first difference, vt, instead of vt, the resulting
test statistics are known as Z and Zt.
13
Haug (1992) provides another set of critical values for the
statistic.
14
These procedures, however, do not allow us to test for the restrictions jointly. Because of
the nonstationarity of the underlying variables the Wald test is also not appropriate. A possible
procedure to use for this purpose is the Johansen (1988) test, but this test has two important
12

60

I. A. MOOSA AND R. H. BHATTI

The results of unit root testing, which are shown in Table 1, are consistent in
indicating that all of the variables underlying equations (4) and (8) are I(1). The
Table 1.Testing for Unit Root ()
Country
Hong Kong

Korea

Malaysia

Philippines

Singapore

Taiwan

Variable

Level

First Difference

st
st+1

1.35
1.04
1.21
1.56
0.35
1.26
0.70
0.20
1.07
0.57
0.58
0.39
0.74
0.46
0.54
1.47
1.19
1.53
0.10
1.10
1.00
0.82
0.99
0.89
2.18
1.51
1.08
0.98
1.42
0.35

7.07*
6.93*
6.99*
7.01*
3.02*
7.34*
7.14*
7.07*
7.20*
4.18*
6.66*
6.52*
6.11*
6.59*
4.66*
7.44*
7.24*
7.44*
7.45*
3.44*
6.16*
6.04*
6.14*
6.01*
4.58*
8.01*
7.47*
7.26*
7.69*
5.32*

qt+1
(p p*)t
st
st+1
qt+1
(p p*)t
st
st+1
qt+1
(p p*)t
st
st+1
qt+1
(p p*)t
st
st+1
qt+1
(p p*)t
st
st+1
qt+1
(p p*)t

Note: * Significant at the 5% level. The 5% critical value is 2.86 (Davidson and MacKinnon,
1993, Table 20.1, p 708).
shortcomings: the difficulty of interpreting multiple cointegrating vectors (likely to arise in the
case of ex ante PPP) and, more importantly, the sensitivity of the results with respect to the
specification of the underlying VARs (the maximum lag, the inclusion of a constant term,
seasonal dummies, etc). Because of these problems we only report the results of the West test.

ARE ASIAN MARKETS INTEGRATED?

61

results of cointegration and coefficient restriction tests, which are presented in Table
2, are highly consistent in lending strong support to UIP and ex ante PPP in all cases,
except for the Philippines in which case the null hypothesis of no cointegration among
the nominal exchange rate, the price ratio and the expected real exchange rate is not
rejected. Moreover, the restrictions implied by UIP (0 = 0 and 1 = 1) and ex ante
PPP ( = 0, = 1, = 1) are not rejected, as judged by the West (1988) corrected t
statistics, in all cases but one (the Philippines).
The results presented in this study on the basis of unconventional model
specifications are more supportive of the two international parity conditions than the
bulk of studies based on conventional specifications. Studies of UIP by, inter alia,
Gaab et al. (1986), Taylor (1987, 1989), and Mayfield and Murphy (1992) produced
results which were generally unsupportive of the hypothesis, unlike the results
produced by Bhatti and Moosa (1995) which indicated that UIP performed reasonably
well in the long run. Similarly, when the PPP model is augmented by the expected
real exchange rate, the results are more supportive of the hypothesis than studies
based on the conventional restricted specification (e.g., Taylor, 1988).
These empirical results must be interpreted with caution and by bearing in mind
some problems associated with the testing techniques and the sample size. The first
problem pertains to the power of unit root and cointegration tests as applied to our
data. The power of these tests is likely to be reduced because of: (i) small-sample
bias, (ii) structural breaks, and (iii) transaction costs acting as thresholds. Small
sample bias arises because the test statistics have asymptotic properties. Monte
Carlo simulations by Banerjee et al. (1986) have shown that small-sample bias is
related to (1 - R2), implying that one should cast doubt on results obtained from
cointegrating regressions having small R 2 (which is not the case here anyway).
Economists have normally circumvented this problem by using data covering a long
period of time and/or using more sophisticated, less conventional techniques
(Taylor, 1995).15 Solving sample-size related problems by using a large sample, if
available, is likely to introduce the problem of structural breaks. With respect to
this study, the possibility of structural breaks could arise because of the measures of
financial deregulation implemented at different points in time which may be
difficult to pinpoint. Given the small sample used in this study, it is not possible to
examine the effect of structural breaks through split samples. However, it is
possible to deal with this problem by using the unit root test proposed by Perron
(1989) which allows for structural breaks whose timing must be known precisely.
Moreover, the Phillips-Ouliaris
statistics can handle heteroscedasticity resulting
from regime changes.
Another relevant problem has been highlighted by Pippenger and Goering (1993),
For example, in testing PPP by detecting mean reversion in the real exchange rate, Abuaf
and Jorion (1990) increased the power of their tests by applying systems estimation (SUR) to a
long time series, while Diebold et al. (1991) employed fractional integration. Both of these
studies produced evidence in favour of PPP.
15

62

I. A. MOOSA AND R. H. BHATTI

ARE ASIAN MARKETS INTEGRATED?

63

64

I. A. MOOSA AND R. H. BHATTI

who examined the power of unit root tests in detecting mean reversion in the presence
of transaction costs. Their conclusion is that the power of unit root tests may fall
dramatically if applied to what they described as threshold processes, where the
threshold is defined by the band representing transaction costs. More specifically,
they concluded that the results of unit root or cointegration tests of economic
relationships implied by arbitrage (e.g., UIP and ex ante PPP) may well be useless in
uncovering the long-run economic relationships between variables in the presence of
transaction cost.16 It is the case, however, that because of the sheer volume of
transactions in the foreign exchange and money markets, the resulting economies of
scale give rise to high transacting efficiency and hence small costs, leading to a
narrow band and a smaller threshold.
What is the relevance of these problems to our results? It must be stressed that the
message here is that these problems reduce the power of the tests to reject the null of
no cointegration and hence to detect market integration. Since the null of no
cointegration is not rejected in one case only (ex ante PPP for the Philippines), this
may be interpreted as implying the strength of market integration which is indicated
by the results of the tests despite the presence of these problems. It could also be the
case that the failure to detect integration between the goods markets of Japan and the
Philippines is due to these problems.
5. CONCLUSIONS
This paper has examined UIP and ex ante PPP as direct tests for the degree of
market integration between Japan and other Asian countries over the period 1980:1 1994:4. The results obtained from residual-based cointegration tests strongly support
the validity of long-run UIP and ex ante PPP in almost all cases. Moreover, the West
(1988) corrected t statistics turned out to be supportive of the coefficient restrictions
implied by the two hypotheses.
The results presented in this study provide stronger evidence for the parity
conditions than what is found in the literature. The positive evidence indicates that
the results of testing international parity conditions are affected not only by the
length of the data sample or the testing technique, but also by model specification.
More importantly, these results are consistent with and confirm the conventional
wisdom that there is a high degree of integration among Asian goods and financial
markets.

16
The basis of this postulation is that the longer the time the process spends inside the
threshold, the longer the time the process behaves as a random walk, reducing the power of the
test in detecting mean reversion. On the other hand, when the process makes the transition
outside the threshold, the faster the speed of adjustment the stronger the signal of mean
reversion which will lead to an increase in the power of the test (p 475).

ARE ASIAN MARKETS INTEGRATED?

65

REFERENCES
Abuaf, N. and Jorion, P., Purchasing Power Parity in the Long Run, Journal of
Finance, March 1990, 157-174.
Adler, M. and Lehman, B., Deviations from Purchasing Power Parity in the Long
Run, Journal of Finance, December 1983, 1471-1487.
Aggarwal, R. and Mougoue, M., Cointegration among Southeast Asian and Japanese
Currencies: Preliminary Evidence of a Yen Bloc?, Economics Letters, February
1993, 161-166.
Banerjee, A., Dolado, J. J., Hendry, D. F., and Smith, G. W., Exploring Equilibrium
Relationships in Econometrics through Static Models: Some Monte Carlo
Evidence, Oxford Bulletin of Economics and Statistics, August 1986, 253-277.
Bhatti, R. H. and Moosa, I. A., A New Approach to Testing ex ante Purchasing
Power Parity, Applied Economics Letters, September 1994, 148-151.
Bhatti, R. H. and Moosa, I. A., An Alternative Approach to Testing Uncovered
Interest Parity, Applied Economics Letters, forthcoming.
Boothe, P., Interest Parity, Cointegration and the Term Structure in Canada and the
United States, Canadian Journal of Economics, August 1991, 595-603.
Bodman, P. M., National Savings and Domestic Investment in the Long Term: Some
Time Series Evidence from the OECD, International Economic Journal, Summer
1995, 37-60.
Davidson, R. and MacKinnon, J. G., Estimation and Inference in Econometrics, New
York, Oxford University Press, 1993.
Dickey, D. A. and Fuller, W. A., Distribution of the Estimators for Autoregressive
Time Series with a Unit Root, Journal of the American Statistical Association,
June 1979, 427-431.
Diebold, F., Husted, S. and Rush, M., Real Exchange Rates under the Gold
Standard, Journal of Political Economy, December 1991, 1252-1271.
Feldstein, M. and Horioka, C., Domestic Saving and International Capital Flows,
Economic Journal, June 1980, 314-329.
Feldstein, M., Domestic Savings and International Capital Movements in the
Long-Run and the Short-Run, European Economic Review, March-April 1983,
139-151.
Frankel, J. A., Measuring International Capital Mobility: a Review, American
Economic Review, Papers and Proceedings, May 1992, 197-202.
Frankel, J. A., Quantifying International Capital Mobility in the 1980s, in Das, D.
K., editor, International Finance: Contemporary Issues, New York, Routledge,
1993.
Gaab, W., Granziol, M. J. and Horner, M., On Some International Parity Conditions:
an Empirical Investigation, European Economic Review, June 1986, 683-713.
Goldstein, M., Mathieson, D. and Lane, T., Determinants and Systematic
Consequences of International Capital Flows, IMF Occasional Papers, No 77,
March 1991.

66

I. A. MOOSA AND R. H. BHATTI

Haug, A., Critical Values for the


Phillips-Ouliaris Test for Cointegration, Oxford
Bulletin of Economics and Statistics, August 1992, 473-480.
Johansen, S., Statistical Analysis of Cointegrating Vectors, Journal of Economic
Dynamics and Control, June-September 1988, 231-254.
Lim, G. C. and Martin, V. L., Regression Based Cointegration Estimators with
Applications, Journal of Economic Studies, Volume 22, Number 1, 1995, 3-22.
Mayfield, E. S. and Murphy, R. G., Interest Rate Parity and the Exchange Risk
Premium: Evidence from Panel Data, Economics Letters, November 1992, 319324.
Modjtahedi, B., An Empirical Investigation into the International Real Interest Rate
Linkages, Canadian Journal of Economics, November 1987, 832-854.
Moosa, I. A., International Parity Conditions as Measures of Regional Market
Integration: a Case Study of Australia and New Zealand, in G. Tower, editor,
Proceedings of the Academy of International Business, Southeast Asia Regional
Conference (Addendum), 1-9, 1995.
Moosa, I. A. and Al-Loughani, N. E., Cross-Sectional Efficiency in the Crude Oil
Market, OPEC Review, Winter 1994, 445-452.
Moosa, I. A. and Bhatti, R. H., The EMS and RIP: Is There Any Connection? Swiss
Journal of Economics and Statistics, forthcoming.
Murphy, R. G., Capital Mobility and the Relationship between Saving and
Investment Rates in OECD Countries, Journal of International Money and
Finance, December 1984, 327-342.
Perron, P., The Great Crash, the Oil Price Shock and the Unit Root Hypothesis,
Econometrica, November 1989, 1361-1401.
Phillips, P. C. B., Time Series Regression with a Unit Root, Econometrica, March
1987, 277-301.
Phillips, P. C. B. and Ouliaris, S., Asymptotic Properties of Residual Based Tests for
Cointegration, Econometrica, January 1990, 165-193.
Pippenger, M. K. and Goering, G. E., A Note on the Empirical Power of Unit Root
Tests under Threshold Processes,Oxford Bulletin of Economics and Statistics,
November 1993, 473-481.
Roll, R., Violations of Purchasing Power Parity and Their Implications for Efficient
International Commodity Markets, in Sarnat, M. and Szego, G. P., editors,
International Finance and Trade, Cambridge, MA: Ballinger, 1979, 133-176.
Taylor, M. P., Risk Premia and Foreign Exchange: a Multiple Time Series Approach
to Testing Uncovered Interest Rate Parity, Weltwirtschaftliches Archiv, Volume
123, Number 4, 1987, 579-591.
Taylor, M. P., An Empirical Examination of Long Run Purchasing Power Parity
Using Cointegration Techniques, Applied Economics, October 1988, 1369-1382.
Taylor, M. P., Vector Autoregressive Tests of Uncovered Interest Rate Parity with
Allowance for Conditional Heteroscedasticity, Scottish Journal of Political
Economy, August 1989, 238-252.
Taylor, M. P., The Economics of Exchange Rates, Journal of Economic Literature,

ARE ASIAN MARKETS INTEGRATED?

67

March 1995, 13-47.


Tobin, J., Domestic Saving and International Capital Movements in the Long Run
and the Short Run by M. Feldstein: Comment, European Economic Review,
March-April 1983, 153-156.
West, K. D., Asymptotic Normality When Regressors Have a Unit Root,
Econometrica, November 1988, 1397-1417.

Mailing Address: Dr. Imad A. Moosa, School of Economics, La Trobe University,


Bundoora, Victoria 3083, AUSTRALIA.
Mailing Address: Dr. Razzaque H. Bhatti, Institute of Economics, University of Azad
Jammu and Kashmir, Muzzaffarabad, PAKISTAN.