International Review of Economics and Finance 16 (2007) 488 502

www.elsevier.com/locate/econbase

A comparison of variance ratio tests of random walk:

A case of Asian emerging stock markets
Hafiz A.A.B. Hoque a , Jae H. Kim b,, Chong Soo Pyun c
a
Department of Finance, University of Dhaka, Bangladesh
b
Department of Econometrics and Business Statistics, Monash University, Australia
c
Fogelman College of Business and Economics, University of Memphis, Memphis, TN 38152, USA
Received 14 June 2005; received in revised form 3 January 2006; accepted 18 January 2006
Available online 6 March 2006

Abstract

This study re-examines the random walk hypothesis for eight emerging equity markets in Asia: Hong Kong, Indonesia,
Korea, Malaysia, the Philippines, Singapore, Taiwan, and Thailand. The hypothesis is tested with two new variance ratio tests
Wright's rank and sign and WhangKim subsampling testsas well as the conventional LoMacKinlay and ChowDenning
tests. We found that (i) the stock prices of the eight Asian countries do not follow random walk with the possible exceptions of
Taiwan and Korea and (ii) the accelerated opening of the eight stock markets to foreign investors following the Asian financial
crisis in 1997 has not significantly altered the mean-reversion patterns of stock price vis--vis relative market efficiency. Our
study affirms that Wright's and WhangKim's tests yield far less ambiguous results as compared to LoMacKinlay and Chow
Denning tests.
2006 Elsevier Inc. All rights reserved.

JEL classification: G14; G15; C14

Keywords: Stock market efficiency; Mean reversion; Variance ratio test; Non-parametric tests; Subsampling

1. Introduction

Since the pioneering work of Lo and MacKinlay (1988), variance ratio (VR) tests have been by far the most widely
used econometric tools for testing the random walk hypothesis (RWH) in emerging equity markets.1 The crux of VR
tests is that if a stock's return is purely random, the variance of k-period return is k times the variance of one-period
return. Hence, the VR, defined as the ratio of 1/k times the variance of the k-period return to the variance of the one-
period return, should be equal to 1 for all values of k.
Chow and Denning (1993) modify LoMacKinlay's VR test so that a set of multiple VRs over a number of holding
periods can be tested to determine whether the multiple VRs are jointly equal to one. Thus, for the purpose of testing a

Corresponding author. Tel.: +61 3 99031596; fax: +61 3 99032007.

E-mail address: Jae.Kim@BusEco.monash.edu.au (J.H. Kim).
1
As shown in Appendix A, 18 published studies are identified as those that dealt with RWH in emerging stock markets. Of the 18, 16 studies used
LoMacKinlay or ChowDenning variance ratio tests along with other tests, such as ARIMA, GARCH, unit root and runs tests.

1059-0560/$ - see front matter 2006 Elsevier Inc. All rights reserved.
doi:10.1016/j.iref.2006.01.001
 H.A.A.B. Hoque et al. / International Review of Economics and Finance 16 (2007) 488502 489

null hypothesis, LoMacKinlay VR test is an individual test while the multiple VR test of Chow and Denning is
a joint test.2 Even though the multiple VR test is quite powerful testing for homoscedastic or heteroscedastic
nulls (Smith, Jefferis, & Ryoo, 2002), it is critical to note that the multiple VR test as well as its forerunnerLo
MacKinlay VR tesare asymptotic tests in that their sampling distributions are approximated by their limiting
distributions.3 Recently, Wright (2000) and Whang and Kim (2003), respectively, have proposed two alternatives
to conventional asymptotic VR testsWright with his rank and sign VR tests and Whang and Kim with their
subsampling version of the Chow and Denning (1993) test. Wright's (2000) tests are exact tests whose
sampling distributions do not resort to asymptotic approximations. Although the validity of Whang and Kim
(2003) test is based on asymptotic arguments, it is important to note that its critical values (p-values) are not
obtained from asymptotic approximation to the sampling distribution: they are obtained from a small sample
approximation.
To the best of our knowledge, Buguk and Brorsen (2003) is the only study that employs Wright's (2000) rank and
sign VR tests in evaluating the RWH for the Istanbul Stock Exchange. Wright's tests have not been applied to Asian
emerging markets; while Whang and Kim's (2003) subsampling test has never been used in the evaluation of the RWH
in any emerging stock markets.
Three motivations for our study are indicated. Firstly, emerging stock markets in Asia have become
increasingly important to the safety and vitality of the global financial markets as shown in recent studies (see
Al-Khazali & Pyun, 2004; Khil & Lee, 2000; Lima & Tabak, 2004). Secondly, since the Asian financial crisis
in 1997, the structure and operations of Asian stock markets have been markedly strengthened and improved,
and there has been a renewed interest on Asian emerging stocks among investors for higher returns through a
global portfolio diversification.4 Lastly, findings on the RWH in Asian emerging markets to date are still
tentative, often fraught with conflicting statistical inferences. For instance, Chang and Ting (2000) show that
while the RWH cannot be rejected for the Taiwan Stock Exchange with monthly, quarterly and annual data, it is
rejected with weekly returns. Huang (1995) reports the rejection of the RWH for the markets of Korea, Malaysia,
Hong Kong, Singapore and Thailand while Ayadi and Pyun (1994), Malliaropulos and Priestley (1999), and
Ryoo and Smith (2002) confirm that the stock prices of these Asian emerging markets are on balance mean-
reverting.5
The purpose of this paper is twofold: (i) to compare inferential outcomes of Wright's and WhangKim VR test
results with those of LoMacKinlay and ChowDenning VR tests for the evaluation of the RWH in eight Asian
emerging equity markets (Hong Kong, Indonesia, South Korea (hereinafter Korea), Malaysia, the Philippines,
Singapore, Taiwan, and Thailand); and (ii) to ascertain any changes in the random walk patterns of stock prices at the
eight individual Asian markets following the Asian crisis with the two new VR tests of Wright (2000) and Whang and
Kim (2003).
Our findings show that the stock prices of the eight emerging markets are mean-reverting, which means that these
markets are not weak-form efficient with the possible exceptions of Taiwan and Korea, and that the accelerated opening
of their respective stock markets to foreign investors in the aftermath of the Asian financial crisis in 1997 has not
significantly altered the random walk patterns of stock price vis--vis market efficiency in the emerging market on
balance. Our findings also support the notion that Wright's rank and sign and WhangKim's subsampling VR tests

2
The Wald test of Richardson and Smith (1991) is an alternative joint VR test. However, we do not consider the Wald test in this study because it
is derived under the assumption that the underlying time series does not exhibit conditional heteroskedasticity. This assumption is too restrictive for
our study, given the evidence of strong conditional heteroskedasticity in stock returns.
3
For example, the limiting distribution of the LoMacKinlay test statistics is the standard normal, and the critical values associated with the
standard normal distribution are used for statistical inference for finite samples.
4
For instance, practically all vestige of ownership restrictions of domestic stocks by foreigners have been lifted literally overnight in Korea and
Thailand under the terms of the International Monetary Fund loans to these countries in 1998. The large influx of foreign portfolio investment into
the two countries followed to take advantage of not only the opening of the markets but also asset prices in these markets, which had been severely
depressed under the crisis environment. The result is that the transaction volume and trading frequency of many stocks in these countries increased
measurably in the recent years.
5
Results are also mixed for emerging markets in Middle East and Africa. El-Erian and Kumar (1995) find that the Turkish market and the
Jordanian market are weak-form inefficient. However, Antoniou and Ergul (1997) find that the Turkish stock market is weak-form efficient only for
stocks with high trading volume.
490 H.A.A.B. Hoque et al. / International Review of Economics and Finance 16 (2007) 488502

yield statistical results, which are less ambiguous as compared to the conventional VR tests, particularly when sample
sizes are relatively small.
The rest of the paper is organized as follows. Section 2 presents a brief literature survey, while Section 3 provides
the data description. Section 4 discusses and compares conventional VR tests with the two new finite sample tests
used in this study. Section 5 reports the empirical results, followed by a summary and our conclusions in the final
section.

2. Brief literature survey

LoMacKinlay's individual VR test reigned the realm of early studies on the RWH in emerging markets. Ayadi and
Pyun (1994) find the Korean equity market meanreverting while Huang (1995) rejects the RWH for the markets of
Korea, Malaysia, Hong Kong, Singapore and Thailand. Interestingly, Chang and Ting (2000) show that the RWH
cannot be rejected for the Taiwan Stock Exchange with monthly, quarterly and annual data, but it is rejected with
weekly returns. Evaluating stock exchanges in China, Darrat and Zhong (2000) and Lee, Chen, and Rui (2001) reject
the RWH. For Latin American stock markets, Urrutia (1995) test the Argentine, Brazil, Chile and Mexico stock
exchanges while Grieb and Reyes (1999) study the Brazil and Mexico exchanges. Both studies find the Latin American
markets are, on balance, weak-form efficient.
There are a number of studies that incorporate ChowDenning's multiple VR tests in the evaluation of RWH in
emerging stock markets. For instance, Kawakatsu and Morey (1999) study 16 sample emerging markets using a battery
of econometric tests. Their results from the ChowDenning tests reveal that the random walk cannot be rejected before
and after financial liberalization for most countries they investigated (p. 406). Ryoo and Smith (2002) find that the
Korean stock market as a whole follows a random walk, and Abraham, Seyyed, and Alsakran (2002) show that the
RWH cannot be rejected for the Kuwaiti, Saudi Arabian, and Bahrain markets when observed index levels are used but
the inferences are reversed with the index data corrected to take into account infrequent trading of small-firm stocks in
the three markets.
Smith et al. (2002) investigate eight emerging stock markets in Africa (Egypt, South Africa, Zimbabwe, Morocco,
Kenya, Nigeria, Botswana, and Mauritius) and find the RHW is rejected for all except the South African market.6
For emerging markets in Europe, Smith and Ryoo (2003) reject the RWH for the stock markets of Greece, Hungary,
Poland, and Portugal, but they find the Turkey market following a random walk. For Asian emerging markets, Lima
and Tabak (2004) report the existence of a random walk for Hong Kong and class A share (available only to
domestic investors) and rejections of RWH for Singapore and class B share (available only to foreign investors) of
the two Chinese stock exchanges. Lastly, Buguk and Brorsen (2003) apply both ChowDenning multiple VR and
Wright's (2000) rank and sign VR tests for the Istanbul stock exchange. They show that the two tests give rise to
conflicting resultsthe multiple VR test finds a random walk, but Wright's nonparametric test shows some
evidence against a random walk.
It should be emphasized that many of the studies cited above augment their VR investigations with additional
methodologies, such as runs test (Urrutia, 1995), unit root test (Kawakatsu & Morey, 1999), ARIMA, GARCH and
artificial neural network (Darrat & Zhong, 2000) and bootstrap test (Lima & Tabak, 2004).

3. Data and summary statistics

We use weekly market prices for the emerging stock markets in eight countries in AsiaHong Kong,
Indonesia, Korea, Malaysia, the Philippines, Singapore, Taiwan, and Thailand. All data are collected from
DataStream. The time period covered is from April 1990 to February 2004. The period yields a total of 726
observations. This provides an excellent window of observing the progress of weak-form market efficiency in the
eight Asian stock markets prior to, and following, the Asian financial crisis. We use Wednesday data. When the
exchanges are closed on Wednesday, Tuesday data are substituted. Log returns for stocks are used throughout the
study.

6
For mixed result for Middle East and Africa, see El-Erian and Kumar (1995), who find that Turkish and the Jordanian markets are not weak-
form efficient, and Antoniou and Ergul (1997), who find that the Turkish stock market is weak-form efficient with a high-trading volume.
 H.A.A.B. Hoque et al. / International Review of Economics and Finance 16 (2007) 488502 491

Table 1
Summary statistics for stock returns in log
HK IND KOR MAL PHI SING TAI THA
Mean 2.73 0.47 1.20 1.44 1.26 0.733 1.35 0.02
S.D. 0.04 0.04 0.05 0.04 0.04 0.03 0.05 0.05
Skewness 0.60 0.08 0.13 0.23 0.14 0.17 0.13 0.07
Kurtosis 4.54 5.39 4.52 10.67 4.51 4.98 4.44 5.69
JB 115.20 173.46 72.13 1782.28 71.70 122.48 65.00 219.60
LM 51.11 61.68 76.31 57.37 55.61 77.32 137.28 59.76
The mean values are multiplied by 103.
JB indicates the JarqueBera test for normality. The test statistics for all countries are significant at the 1% level.
LM indicates the Lagrange multiplier test for conditional heteroskedasticity (ARCH) with 10 lags. The statistics for all countries are significant at the
1% level.

Table 1 presents summary statistics for the stock returns. It is evident from the JarqueBera test for normality that
the returns are non-normal, while the Lagrange multiplier (LM) test for the presence of the ARCH effect indicates
clearly that all stock returns show strong conditional heteroskedasticty. Fig. 1 presents the time plots of all time series in
natural logarithm. Not surprisingly, all time series show distinctive downward spikes around 1997, corresponding to
the timing of the Asian financial crisis. To isolate the effect of the Asian crisis on the outcomes of the variance ratio
tests, we divide the data into two parts. The first sub-period covers the period from 1990 to 1996 with 352 observations
while the second sub-period covers from 1998 to 2004 with 321 observations. The variance ratio tests are applied to the
whole sample as well as to these sub-periods.

9.2 5.6 6.0

8.8 5.8
5.2
8.4 5.6

4.8 5.4
8.0
5.2
7.6 4.4 5.0
7.2 4.8
4.0
6.8 4.6
6.4 3.6 4.4
1990 1992 1994 1996 1998 2000 2002 1990 1992 1994 1996 1998 2000 2002 1990 1992 1994 1996 1998 2000 2002
HONGKONG INDONESIA KOREA

7.2 7.2 6.8

7.0 6.6
6.8
6.8
6.4
6.4
6.6
6.2
6.4 6.0
6.0
6.2
5.6
6.0 5.8
5.2 5.6
5.8
5.6 4.8 5.4
1990 1992 1994 1996 1998 2000 2002 1990 1992 1994 1996 1998 2000 2002 1990 1992 1994 1996 1998 2000 2002
MALAYSIA PHILIPPINES SINGAPORE

6.0 7.6

5.6 7.2

5.2 6.8

4.8 6.4

4.4 6.0

4.0 5.6
1990 1992 1994 1996 1998 2000 2002 1990 1992 1994 1996 1998 2000 2002
TAIWAN THAILAND

Fig. 1. Time plots of stock indices (in natural logs). Note: The shaded area in each plot indicates the period of the Asian crisis.
492 H.A.A.B. Hoque et al. / International Review of Economics and Finance 16 (2007) 488502

4. Methodologies

4.1. LoMacKinlay test

Central to the evaluation of the RWH is a mean-reverting pattern of stock returns. If the price of a stock is mean-
reverting, its return is predictable ex ante in the form of a systematic pattern in its dependence on past prices. On the
other hand, if the stock price follows a random walk or martingale, the stock return is unpredictable from past price
information.
Suppose that xt is an asset return at time t, where t = 1,, T. Following Wright (2000), we write
( ) ( )
1 X T  2 1X T  2
VR x; k xt xt1 N xtk1 kl  xt l ; 1
Tk tk T t1
P
where l T 1 Tt1 xt . This is an estimator for the unknown population VR, denoted as V(k), which is the ratio of 1/k
times the variance of the k-period return to the variance of the one-period return. Lo and MacKinlay (1988) showed that
if xt is independent and identically distributed (iid), then under the null hypothesis that V(k) = 1,
 
22k1k1 1=2
M1 x; k VR x; k 1 2
3kT
follows the standard normal distribution asymptotically. To accommodate xt's exhibiting conditional heteroscedas-
ticity, Lo and MacKinlay (1988) proposed a heteroscedasticity robust test statistic
k1 
X  !1=2
2kj 2
M2 x; k VR x; k 1 dj 3
j1
k
which follows the standard normal distribution asymptotically under null hypothesis that V(k) = 1, where
( ) 
XT hXT i2
2 2 2
dj xt l xtj l  x l
t1 t
:
tj1

The M2 test is applicable to xt's generated from a martingale difference time series (see Assumption H of Lo and
MacKinlay, 1988). The usual decision rule for the standard normal distribution applies to both tests.

4.2. ChowDenning test

The LoMacKinlay test is an individual test where the null hypothesis is tested for an individual value of k. The
question as to whether or not a stock return is mean-reverting requires that the null hypothesis hold true for all values of
k. In view of this, it is necessary to conduct a joint test where a multiple comparison of VRs over a set of different time
horizons is made. However, conducting separate individual tests for a number of k values may be misleading as it tends
to over-reject the null hypothesis of a joint test. Thus, the weakness of LoMacKinlay's test is that it ignores the joint
nature of testing for the RWH. It may involve much larger Type I error than the nominal level of significance. Namely,
the probability of incorrect rejection of the true null hypothesis can be quite larger than the chosen level of significance
(see Savin, 1984 for details). To avoid this problem, ChowDenning (1993) devise a joint test with controlled size as
follows.
Under the null hypothesis, V(ki) = 1 for i = 1, , l against the alternative hypothesis that V(ki) 1 for some i. Chow
and Denning's (1993) test statistic is
p
MV1 T max jM1 x; ki j; 4
1ViVl
where M1(x;ki) is defined in (2). This is based on the idea that the decision regarding the null hypothesis can be made
based on the maximum absolute value of the individual VR statistics.7 The null hypothesis is rejected at level of

7
The statistic follows the studentized maximum modulus distribution with l and T degrees of freedom. The critical values of the test are tabulated
in Hahn and Hendrickson (1971) and Stoline and Ury (1979). It should be noted that when T is large, the critical values of the test can be calculated
based on the limiting distribution of the statistic.
 H.A.A.B. Hoque et al. / International Review of Economics and Finance 16 (2007) 488502 493

significance if the MV1 statistic is greater than the [1 ( / 2)]th percentile of the standard normal distribution
where = 1 (1 )1/l. Similarly, the heteroskedasticityrobust version of the ChowDenning test MV2 can be
written as
p
MV2 T max jM2 x; ki j; 5
1ViVl

which is a joint test using M2(x,k) given in (3), and it has the same critical values as MV1.

4.3. Wright's rank and sign VR tests

As already noted, both LoMacKinlay and ChowDenning tests are asymptotic tests, whose sampling
distributions are approximated based on their limiting distributions. Lo and MacKinlay (1989) find that the sampling
distribution of the VR statistic can be far from normal in finite samples, showing severe bias and right skew. These
finite sample deficiencies may give rise to serious size distortions or low power, which can lead to misleading
inferences. This is especially true when the sample size is not large enough to justify asymptotic approximations. As
Deo and Richardson (2003) point out, the LoMacKinlay test is inconsistent with respect to the variety of mean-
reverting alternatives where the limiting power function is bounded by a number less than one. In this respect,
Wright's (2000) tests have two advantages over LoMacKinlay and ChowDenning tests when sample size is
relatively small: (i) as the sign and rank tests have exact sampling distribution, there is no need to resort to
asymptotic approximation and (ii) the tests may be more powerful than the conventional VR tests when the data are
highly non-normal (Wright, 2000).
Wright (2000) derives rank and sign statistics as follows. Let r(xt) be the rank of xt among xt's. Consider the
standardized rank r1t = [(r(xt) 0.5(T + 1) / [(T 1)(T + 1) / 12]. Under the null hypothesis that xt is generated from an i.i.
d. sequence, r(xt) is a random permutation of the numbers of 1, , T with equal probability. Wright (2000) suggests the
statistic
0 1
1 P
T
C22k1k11=2
2
B Tk r 1t r 1t1 N r 1tk1
B tk C
R1 B 1C ; 6
@ PT A 3kT
T 1 r1t 2
t1

which follows an exact sampling distribution (see Proposition 1 of Wright, 2000). Wright proposes the use of
alternative standardization r2t = 1[r(xt) / (T + 1)], where is the standard normal cumulative distribution function.
This leads to the R2 statistic, which can be written as follows:
0 1
1 P
T
C22k1k11=2
2
BTk r2t r2t1 N r2tk1
B tk C
R2 B 1 C : 7
@ 1
P
T
2
A 3kT
T r2t
t1

The R2 test in (7) shares the same sampling distribution as R1. The critical values of these tests can be obtained by
simulating their exact distributions, as shown in Table 1 of Wright (2000).
Similarly, Wright (2000) derives a sign-based test statistic. Consider st = 2u(xt,0) and u(xt,0) = 1(xt > 0) 0.5 while 1
(.) is the indicator function which takes the value 1 if the condition inside the bracket is satisfied and 0 otherwise. Under
the null hypothesis that xt is a martingale difference sequence (Assumptions A1 and A2 of Wright, 2000) whose
unconditional mean is zero, st is an i.i.d. sequence with mean 0 and variance equal to 1, which takes the value 1 and 1
with equal probability of 0.5. Based on this, Wright (2000) proposed the sign-based test statistic
0 1
1 P
T
2
BTk st st1 N stk1 C22k1k11=2
B tk C
S1 B 1 C : 8
@ 1
P
T
2
A 3kT
T st
t1
494 H.A.A.B. Hoque et al. / International Review of Economics and Finance 16 (2007) 488502

Similarly to R1 and R2 tests, the critical values of the S1 test can be obtained by simulating its exact sampling
distribution (see Proposition 2 of Wright, 2000). It should be mentioned that Wright's (2000) S2 test is not considered
here, as his Monte Carlo simulation results clearly indicate that its size and power properties are quite inferior to those
of S1. The critical values of the R1, R2 and S1 tests associated with the sample sizes and holding periods are given in
Table 2.

4.4. WhangKim subsampling test

The WhangKim test uses the subsampling technique of Politis, Romano, and Wolf (1997), which is a data-intensive
method of approximating the sampling distribution. It can show better properties than the conventional VR tests when
the sample size is relatively small. The Monte Carlo experiment results reported in Whang and Kim (2003) confirm that
their new VR test shows excellent power in small samples, coupled with little or no serious size distortions.
To test the null hypothesis that V(ki) = 1 (i = 1, , l), Whang and Kim (2003) consider the statistic
p
MV3 T gN x1 ; N ; xT ; 9
where gt x1 ; N ; xT max1 V i V l jVRx; ki 1j and VR(x;k) is as defined in (1). The sampling distribution function
for the MV3 statistic is written as
p 
GT x P T gT x1 ; N ; xT Vx : 10

Since the distribution function given in (10) is unknown and analytically intractable, Whang and Kim (2003) use the
following approximation. Consider a subsample (xt, , xtb+1) of size b for t = 1, , T b + 1. The statistic MV3
calculated from the subsample is denoted as gT,b,t = gb(xt, , xtb+1). Then, GT(x) is approximated by the distribution
function obtained by the collection of gT,b,t's calculated from all individual subsamples. It can be written as
TX
b1  p 
GT ;b x T b 21 1 bgT ;b;t V x ;
t0

where 1(.) is the indicator function that takes 1 if the condition inside the bracket is satisfied and 0 otherwise.

Table 2
Critical values for Wright's R1, R2 and S1
k T = 350 T = 750
5% 10% 5% 10%
R1
2 2.04, 1.83 1.73, 1.50 2.02, 1.90 1.70, 1.59
4 2.03, 1.82 1.75, 1.49 1.99, 1.88 1.72, 1.55
8 1.98, 1.82 1.75, 1.42 1.95, 1.89 1.70, 1.51
16 1.93, 1.67 1.72, 1.32 1.93, 1.82 1.71, 1.44

R2
2 2.07, 1.82 1.74, 1.50 2.04, 1.89 1.72, 1.58
4 2.02, 1.82 1.74, 1.51 1.97, 1.85 1.70, 1.55
8 1.98, 1.82 1.73. 1.47 1.94, 1.85 1.69, 1.53
16 1.94, 1.69 1.73, 1.33 1.93, 1.76 1.71, 1.45

S1
2 2.03, 1.92 1.71, 1.60 1.97, 1.97 1.68, 1.61
4 1.97, 1.94 1.69, 1.63 1.99, 1.97 1.68, 1.64
8 1.91, 1.97 1.65, 1.59 1.88, 1.99 1.62. 1.65
16 1.83, 1.94 1.63, 1.53 1.88, 1.98 1.64, 1.62
The critical values were simulated with 10,000 replications in each case.
The 5% (10%) critical values represent the 2.5th (5th) and 97.5th (95th) percentiles of the simulated distributions of the statistic.
 H.A.A.B. Hoque et al. / International Review of Economics and Finance 16 (2007) 488502 495

The 100(1 )% critical value for the test can be calculated as the (1 )th percentile of b,T, while the p-value of
the test is estimated as 1 T,b(MV3). The null hypothesis that V(ki) = 1 (i = 1, , l) is rejected at the level of significance
if the observed MV3 is greater than this critical value or if the p-value is less than . To implement the subsampling
technique, a choice of block length b should be made. Whang and Kim (2003) recommend that a number of block
lengths from an equally spaced grid in the interval of [2.5T 0.3, 3.5T 0.6] be taken. However, they find that the size and
power properties of their test are not sensitive to the choice of block length.8

5. Empirical findings

Table 3 reports the individual test statistics (R1, R2, S1, and M2) for the whole sample.9 In making inferential
decisions using these statistics, we reject the null of RWH (or martingale) if there are more than two rejections
at any of three levels of significance we have considered (1%, 5% and 10%).10 For example, the case of Hong
Kong in Table 5 shows two rejections at 10% level using R2 test. Our overall decision for this case is the
rejection of the RWH hypothesis. According to Wright's (2000) findings, the R1 and R2 tests are as powerful
as S1.
It is apparent that the null of RWH is rejected for the eight individual emerging markets at the 5% level
of significance. However, Korea shows weaker evidence against the RWH than those of the other markets,
as only the S1 statistic of Korea shows rejection when k is 2 and 4. It should be noted that all test
statistics (R1, R2, S1, and M2) reported for the eight individual country markets give rise to fairly consistent
inferences with the exception of Malaysia, whose RWH is rejected by R1, R2, and S1 tests, but not by M2
tests.
The same divergence amongst the four test statistics is observed for Malaysia again in Table 5. In other
words, RWH is rejected for the Kuala Lumpur Stock Exchange for the whole sample period between 1990 and
2004 as well as the sub-sample period covering the post-Asian currency crisis period between 1998 and 2004
when the hypothesis evaluated by R1, R2, and S1 tests. However, M2 test does not reject for the same two time
periods.11
Table 4 presents the test results for sub-period I, covering from 1990 to 1997. Based on the four test
statistics (R1, R2, S1, and M2) evaluated, the RWH is rejected for all countries, except Korea, Taiwan and
Thailand. An incongruence among the four test statistics is observed for only one country, Singapore, whose
RWH is rejected by R1, R2, and S1 tests, but not by M2 test. While the sources for this incongruence is
difficult to ascertain like those for Malaysia, we reject the RWH for Singapore as it is likely that Wright's R1,
R2, and S1 tests capture the relevance of serial dependence in past prices more effectively than the M2 test does
in this case, as the former are more powerful than the latter according to the simulation results of Wright
(2000).
Table 5 presents the test results for sub-period II between 1998 and 2004, which covers the post-Asian financial
crisis period. The null hypothesis is rejected for all countries except for Korea and Taiwan, where the evidence against

8
As an alternative to the subsampling technique, one may suggest the use of the bootstrap method of Efron and Tibshirani (1993) to approximate
the sampling distribution of the VR test statistic, as in Malliaropulos and Priestley (1999). However, theoretical validity of bootstrapping the VR test
has yet to be established, and its small sample properties are also unknown. Given these unknown properties, the bootstrap test is left for future
research.
9
We decide not to report the M1 statistics because it is well known that the M1 test shows small sample properties far inferior to M2 when
the underlying time series shows conditional heteroskedasticity. By the same reason, we report only the MV2 statistics for the ChowDenning
test.
10
This particular decision rule is adopted to make the testing procedure operationally easier. However, it should be noted that this decision rule
may render the test severely over-sized, although it gives higher power (see Section 4.2 for the related discussion). We are indebted to an
anonymous referee for the clarification of this point.
11
Since the four test statistics (R1, R2, S1, and M2) give consistent inference for the sub-sample test covering the pre-Asian currency crisis
period between 1990 and 1997, the sources for the incongruence among the four test statistics for the 19902004 and 19982004 may be
attributable to two factors. First, the distortion in the stock price indices induced by Malaysia's capital control which embargoed foreign
currency movement following the Asian currency crisis in 1997. During the embargo period foreign investors curtailed their buying and
selling of stocks listed on the Kuala Lumpur Stock Exchange (see Perkins and Woo, 1998). Second, since the R1, R2, and S1 tests have higher
power than the M2 test as evident in the simulation results of Wright (2000), the former may have captured the serial dependence of past
prices better.
496 H.A.A.B. Hoque et al. / International Review of Economics and Finance 16 (2007) 488502

Table 3
Individual test results (whole sample)
k R1 R2 S1 M2
#
Hong Kong 2 1.18 1.09 1.89 0.85
4 2.15 2.42 2.46 2.21
8 1.62# 1.92 2.57 1.93#
16 1.03 1.15 3.03 1.40
Indonesia 2 3.76 3.58 3.53 2.36
4 5.25 5.47 4.55 4.11
8 5.69 6.12 4.16 4.72
16 3.97 3.98 2.69 2.89
Korea 2 0.95 0.24 2.49 0.74
4 1.34 0.75 1.63# 0.33
8 1.50 1.41 0.86 1.04
16 1.15 1.23 0.06 0.99
Malaysia 2 2.66 1.98 2.41 0.26
4 2.60 2.10 2.16 0.20
8 3.73 3.35 2.63 0.83
16 4.39 3.90 3.16 1.16
Philippines 2 2.88 3.14 1.67# 2.56
4 4.45 5.00 2.84 4.22
8 4.18 4.86 2.70 4.30
16 3.89 4.44 2.53 4.02
Singapore 2 2.16 2.12 2.19 1.70#
4 2.28 2.32 2.22 1.88#
8 2.19 2.27 2.21 1.94#
16 2.62 2.65 2.21 2.18
Taiwan 2 1.12 1.28 0.63 0.87
4 1.40 1.65# 0.12 1.33
8 1.45 1.85 0.65 1.61
16 1.47# 1.58# 1.40# 1.20
Thailand 2 1.98 2.05 1.52 1.72#
4 2.81 2.99 1.63# 2.44
8 2.87 2.85 1.93 2.25
16 2.63 2.25 2.13 1.50
and # indicate significance at the 5% and the 10% levels, respectively.

the null is fairly weak. Overall, it can be inferred from individual tests that since 1990 the stock markets in Hong Kong,
Indonesia, Malaysia, the Philippines, Singapore and Thailand have not been weak-form efficient, while those in Korea
and Taiwan have been. Our findings also suggest that the Thailand equity market was weak-form efficient in the 1990s
leading up to the Asian crisis in 1997.
Our findings regarding the Korean and Thailand markets are most interesting. Both countries bore the brunt of
the devastating impacts of the Asian financial crisis, and both countries were forced to accelerate the opening of
their stock markets to foreign investors under the mandate of the International Monetary Fund conditional lending
in 19971998. Yet, the weak-form efficiency of the Korean stock market appears on balance unaffected by the
further opening of its equity market following the Asian financial crisis, while that of the Thailand market seems to
have drifted toward relative inefficiency, especially during the recovery stage of its market following the Asian
crisis.
Table 6 reports the results from the joint test results. It is of particular interest to see how Wright's test
results are compared to the joint test results. To make the comparisons of the test results stand out, the
overall outcomes of Wright's tests are given in the last column of Table 6. For the whole sample period, the
WhangKim test shows rejection of RWH for all eight individual countries, as the country's p-values are less
than 5% for nearly all cases evaluated. These findings are comparable to the results we obtained from
Wright's tests.
For sub-period I (between 1990 and 1997), there are three cases where the joint and individual tests give rise to
conflicting inferences. According to the WhangKim test, the null hypothesis of random walk cannot be rejected for
 H.A.A.B. Hoque et al. / International Review of Economics and Finance 16 (2007) 488502 497

Table 4
Individual test results (sub-period I: 19901997)
k R1 R2 S1 M2
Hong Kong 2 0.94 0.47 2.30 0.23
4 1.24 1.07 2.74 0.63
8 0.49 0.55 2.81 0.44
16 0.14 0.20 3.41 0.20
Indonesia 2 5.37 5.63 3.79 4.24
4 6.12 6.28 4.48 4.99
8 5.35 5.47 3.74 4.79
16 5.16 5.08 3.37 4.49
Korea 2 0.93 0.39 1.87 0.19
4 1.36 0.90 1.48 0.34
8 1.24 0.87 0.61 0.33
16 0.70 0.51 0.19 0.12
Malaysia 2 1.26 0.56 1.65# 0.23
4 1.20 0.76 1.43 0.16
8 1.19 0.72 1.80# 0.36
16 1.46# 1.10 2.30 0.01
Philippines 2 2.82 3.45 2.19 3.04
4 3.37 3.88 2.85 3.32
8 2.81 3.48 2.43 3.18
16 1.88 2.57 1.81# 2.56
Singapore 2 2.53 2.39 2.40 1.60
4 2.64 2.57 2.71 1.70#
8 2.03 1.98 2.56 1.52
16 1.85 1.98 2.19 1.56
Taiwan 2 0.53 0.83 0.05 0.76
4 0.64 1.07 0.71 1.07
8 0.74 1.36 0.74 1.43
16 0.42 0.83 0.72 0.82
Thailand 2 0.98 0.63 1.23 0.54
4 0.68 0.40 0.60 0.31
8 1.15 1.00 0.91 0.81
16 1.20 1.08 0.82 0.80
and # indicate significance at the 5% and the 10% levels, respectively.
Sub-period I covers the period from 1990 to 1997 with 352 observations.

the stock markets in Hong Kong and Malaysia. For the Taiwanese market, the WhangKim test rejects the hypothesis,
while the Wright's test yields otherwise. It should be noted that, at the 5% level of significance, neither the Whang
Kim nor Wright's test reject the RWH for Thailand. For sub-period II (between 1998 and 2004), the WhangKim and
Wright's tests provide similar outcomes for all markets, except for Hong Kong, which the WhangKim test find as
weak-form efficient.
The following conclusions can be drawn from the overall results. On balance, stock returns in the eight Asian
emerging markets are found not to follow the random walk, indicating that these markets have not been weak-form
efficient. There are two exceptions: the Korean market, which is found relatively weak-form efficient both before and
after the Asian financial crisis in 1997, and the Taiwanese market, which is found weak-form efficient only during the
post-Asian crisis period.
The findings on the Thailand and Hong Kong markets are intriguing. Thailand's market efficiency is found as
declining in the aftermath of the Asian crisis, while the Hong Kong market needs further investigation because the
findings from the joint and individual tests are contradictory. For all other markets under studyIndonesia,
Malaysia, the Philippines, and Singaporethere is little evidence to suggest that the markets are weak-form
efficient.
It is instructive to compare our findings with those of other studies, which tackled the same issue using the
conventional LoMacKinlay or ChowDenning VR tests. Our findings regarding the markets in Korea and
Taiwan are generally consistent with those reported in Ayadi and Pyun (1994) and Ryoo and Smith (2002), both
498 H.A.A.B. Hoque et al. / International Review of Economics and Finance 16 (2007) 488502

Table 5
Individual test results (sub-period II: 19982004)
k R1 R2 S1 M2
Hong Kong 2 0.46 0.35 0.11 0.36
4 1.36 1.55# 0.27 1.47
8 1.28 1.51# 0.51 1.42
16 0.58 0.53 0.55 0.53
Indonesia 2 1.30 1.66# 1.34 1.47
4 1.99 2.59 1.34 2.39
8 2.87 3.52 1.20 3.14
16 1.06 1.40 0.56 1.37
Korea 2 0.56 0.23 2.01 0.35
4 0.37 0.17 1.11 0.15
8 0.64 0.90 0.68 0.85
16 0.67 1.01 0.25 0.97
Malaysia 2 2.96 2.78 1.90 1.06
4 3.12 3.24 1.88 1.44
8 3.99 4.12 1.95 1.74#
16 4.20 4.05 1.96 1.66#
Philippines 2 1.63# 1.79# 0.56 1.69#
4 2.97 3.54 1.43 3.21
8 2.63 3.17 1.37 2.95
16 2.53 2.72 1.62 2.42
Singapore 2 1.28 1.39 1.01 1.48
4 1.37 1.62# 0.57 1.64
8 1.47# 1.79# 0.64 1.92#
16 1.79# 1.88 1.03 2.02
Taiwan 2 0.93 0.69 0.67 0.50
4 1.15 1.03 0.66 0.79
8 0.85 0.71 0.96 0.52
16 1.32 1.12 1.84# 0.85
Thailand 2 1.96 1.87 1.23 1.38
4 2.93 3.00 1.61# 2.39
8 2.57 2.40 1.64# 1.77#
16 2.37 2.01 2.07 1.41
and # indicate significance at the 5% and the 10% levels, respectively.
Sub-period II covers the period from 1998 to 2004 with 321 observations.

of which confirm that the stock prices of these two Asian emerging markets are on balance weak-form efficient.
Our findings regarding the markets in Malaysia, Singapore and Thailand are also in agreement with Huang
(1995), who rejects the RWH for these three markets in the early 1990s. However, it is interesting to note that
Huang's (1995) study also rejects the RWH for the Korean and Hong Kong stock markets in contrast to our
findings, that the Korean market is relatively weak-form efficient while the relative efficiency of the Hong Kong
market is indeterminable. More significantly, Chang and Ting (2000) reject the RWH for the Taiwanese market in
contrast to our finding that the Taiwanese market is weak-form efficient during the post-Asian financial crisis
period.
It should be noted that there are many cases, even in our own study, where the conventional VR tests give rise to
inferences that are very different from what we obtained with Wright's and Whang and Kim's tests, respectively. The
whole sample result from Table 6 serves as a good example. We see that, at the 5% level of significance, the Chow
Denning test does not reject the RWH for the stock markets of Hong Kong, Korea, Malaysia, Singapore and Thailand,
while the WhangKim test decisively does for these countries. Similar contradictory results are found in individual
tests as well. From Table 3, the M2 test does not reject the RWH for Malaysian and Taiwanese markets, but Wright's test
strongly rejects them.
Given the improved size and power properties of Wright's (2000) and Whang and Kim's (2003) tests, our
overall findings strongly suggest that the empirical validity of weak-form efficiencies of emerging stock markets,
including those of non-Asian emerging economies, based on the conventional VR test, should be re-examined. This
 H.A.A.B. Hoque et al. / International Review of Economics and Finance 16 (2007) 488502 499

Table 6
Joint test results
ChowDenning MV2 WhangKim test Wright
Whole sample
Block Lengths
41 64 87 110 133 156
HK 2.21 0.00 0.00 0.00 0.00 0.00 0.02 Reject
IND 4.72 0.00 0.00 0.00 0.00 0.01 0.02 Reject
KOR 1.04 0.16 0.06 0.03 0.01 0.01 0.01 Reject
MAL 1.16 0.00 0.00 0.01 0.01 0.02 0.03 Reject
PHI 4.30 0.00 0.00 0.00 0.00 0.00 0.00 Reject
SIN 2.18 0.00 0.00 0.00 0.00 0.00 0.00 Reject
TAI 1.61 0.00 0.00 0.00 0.03 0.06 0.08 Reject
THA 2.44# 0.01 0.01 0.00 0.00 0.00 0.00 Reject

Sub-period I
Block Length
28 42 56 70 84 98
HK 0.63 1.00 0.92 0.89 0.87 0.88 0.84 Reject
IND 4.99 0.00 0.00 0.00 0.00 0.00 0.00 Reject
KOR 0.34 1.00 0.96 0.89 0.87 0.90 0.95 Accept
MAL 0.36 1.00 0.88 0.72 0.75 0.67 0.65 Reject
PHI 3.32 0.00 0.00 0.00 0.00 0.00 0.00 Reject
SIN 1.70 0.00 0.00 0.00 0.00 0.00 0.00 Reject
TAI 1.43 0.00 0.00 0.02 0.06 0.07 0.13 Accept
THA 0.81 0.37 0.09 0.07 0.12 0.08 0.06 Accept

Sub-period II
Block Length
41 64 87 110 133 156
HK 1.47 0.15 0.27 0.21 0.25 0.17 0.19 Reject
IND 3.14 0.00 0.00 0.00 0.02 0.02 0.01 Reject
KOR 0.97 0.40 0.38 0.39 0.20 0.07 0.12 Accept
MAL 1.74 0.00 0.00 0.00 0.00 0.00 0.00 Reject
PHI 3.21 0.00 0.00 0.00 0.00 0.00 0.00 Reject
SIN 2.02 0.00 0.00 0.00 0.00 0.00 0.00 Reject
TAI 0.85 0.30 0.20 0.12 0.12 0.15 0.16 Accept
THA 2.39# 0.00 0.03 0.03 0.00 0.02 0.02 Reject
The 5% and 10% critical values for MV2 statistic are 2.49 and 2.23, respectively.
For all cases, the values of k used are 2, 4, 8, 16.
Following Whang and Kim (2003), block lengths are chosen from an equally spaced grid in the interval [2.5T 0.3, 3.5T 0.6].
Sub-period I covers the period from 1990 to 1997 with 352 observations.
Sub-period II covers the period from 1998 to 2004 with 321 observations.
The Wright column presents the overall results from Wright's rank and sign test results: Accept denotes overall acceptance of the null of random
walk and Rejection overall rejection of the null. The guideline as to how these decisions are made is given at the beginning of Section 5.

is because the widely used LoMacKinlay and ChowDenning tests may have rendered misleading statistical
inferences.

6. Concluding remarks

In this paper, we implemented alternative variance ratio tests to evaluate the issue of whether or not emerging stock
markets follow a random walk: were stock prices in the eight emerging markets in Asia mean-reverting? The
conventional tests implemented are the LoMacKinlay and the ChowDenning tests, while Wright's (2000) rank and
500 H.A.A.B. Hoque et al. / International Review of Economics and Finance 16 (2007) 488502

sign tests and Whang and Kim's (2003) subsampling test are fairly new. The main advantage of the latter over the
former is that the latter does not rely on the asymptotic approximations to the sampling distributions of the statistic.
This is because Wright's and WhangKim's methods use the exact distributions or data-intensive approximations to
the sampling distribution of the statistic. Monte Carlo simulation results of these two tests, reported in Wright (2000)
and Whang and Kim (2003), clearly show that the two new finite sample tests give rise to better size and power
properties in their critical statistics.
The major findings of our study are that during the period under study (from 1990 to 2004) five individual stock
markets in emerging countriesIndonesia, Malaysia, the Philippines, Singapore and Thailandshow significant
mean-reverting and predictable behavior in their weekly returns series, while two marketsTaiwan and Korea
exhibit some mean-reverting, but largely unpredictable patterns in the same series. These findings are in large part
consistent with many recent studies,12 which show that the stock price behavior of many emerging markets are
often inter-temporally predictable and that the pattern of their inter-temporal predictability can be exploited by
astute investors.
While it is more pronounced in emerging markets, the inter-temporal predictability of mean-reversion in
stocks is also observed in the U.S. and European markets. The crux of the issue within the framework of the
efficient market hypothesis is whether or not stock markets capture sufficient risk premium into future stock
prices so that investors relying on past price information will be able to earn risk-adjusted abnormal returns.
Even though it is generally agreed that equity markets in developed countries are weak-form efficient, the
phenomena of inter-temporal predictability have been reported for both developed and emerging stock markets
and the phenomena are still defined and implemented in the financial literature largely as empirical
questions.13
It should be emphasized that this predictability of mean-reversion found in the Asian emerging markets may be
exacerbated by two structural factors common to all emerging equity markets. Firstly, the markets may suffer from
the effect of thinly traded stocks: not only are the sizes of many listed firms small, but also shares of these firms are
thinly traded at times. Abraham et al. (2002) on Middle East stock markets show that when raw data are used, the
RWH is rejected. When the data are corrected for thinly traded stocks, statistical inference is opposite to what they
had found with the uncorrected data. Secondly, the effects of information asymmetry among various participants are
widespread (Hahm & Mishkin, 2000) and their diverse financial institutions are still in an early stage of growth. In
this respect, the mean-reverting patterns detected in many Asian emerging markets may strongly suggest inter-
temporal inefficiency, which may be more structural in its origin than occasional, irrational behavior on the part of
investors. In this respect, our results reinforce the findings by Van der Hart, Stagter, and van Dijik (2003) that inter-
temporal inefficiency in emerging markets does exist and that the inefficiency may be exploitable by both domestic
and foreign investors.
Finally, our analysis with Asian data shows that Wright's (2000) rank and sign and Whang and Kim's (2003)
subsampling tests yield inferential outcomes that are far less ambiguous as compared to the LoMacKinlay and Chow
Denning VR tests, particularly when sample size is small. We believe that the two new VR tests (Wright's and Whang
Kim's tests) constitute an experimental tool of sufficient quality worthy for the re-evaluation of the RWH reported in
the literature on other emerging markets of the world.

Acknowledgements

We would like to thank the Journal editor (Carl H. Chen) and two anonymous referees for their helpful comments.
An earlier version of this paper was presented at the annual joint meeting of the Korean Finance Association and
KoreaAmerica Finance Association in Dogok, Korea in 2003. This study was partly supported by a summer research
grant from College of Business and Center for International Business Education and Research at the University of
Memphis.

12
See Smith and Ryoo (2003), Buguk and Brorsen (2003), Nam, Pyun, and Kim (2003), Lima and Tabak (2004) and Seddighi and Nian (2004).
13
In the literature, the issue is being debated in terms of investor behaviors, focusing on their rational vs. irrational expectations that have given
rise to the so-called momentum and contrarian investment strategies. An excellent survey article on the subject is found in Malkiel (2003) (see
also Slezak, 2003). We are indebted to an anonymous referee for pointing out this significant issue.
 H.A.A.B. Hoque et al. / International Review of Economics and Finance 16 (2007) 488502 501

Appendix A

List of selected RWH studies and methodologies used

Studies Countries covered Methodologies used Sample frequency and range
Ayadi and Pyun (1994) Korea LoMacKinlay VR D, W, M, 2M, 1984:11998:12
Huang (1995) 8 Asian countries, plus Japan LoMacKinlay VR, Unit root W, 1988:11992:6
Urrutia (1995) Argentina, Brazil, Chile Mexico LoMacKinlay VR, Runs W, 1980:31988:12
Laurence, Cain, and Qian (1997) China Unit root D, 1993:31996:10
Grieb and Reyes (1999) Brazil, Mexico LoMacKinlay VR W, 1988:121995:6
Kawakatsu and Morey (1999) 16 emerging markets worldwide Unit root tests, ChowDenning MVR M, 1976:11997:12
Malliaropulos and Priestley (1999) Southeast Asian markets LoMacKinlay, Bootstrap W, 1988:11995:12
Chang and Ting (2000) Taiwan LoMacKinlay VR, W, M, Q, Y, 1971:11996:1
Darrat and Zhong (2000) China LoMacKinlay VR,ARIMA,GARCH D, 1990:121998:10
Lee et al. (2001) China LoMacKinlay VR, Unit root, D, 1990:121997:12
Artificial neural networks
Abraham et al. (2002) Kuwait, Saudi Arabia, Bahrain LoMacKinlay VR, Runs W, 1992:101998:12
Ryoo and Smith (2002) Korea ChowDenning MVR D, 1988:31998:12
Smith et al. (2002) 15 African countries ChowDenning MVR W, 1990:11998:8
Chang, Lima, and Tabak (2003) 11emerging markets plus U.S LoMacKinlay VR, Variable moving D, 1992:12002:12
and Japan average and technical trading rules
Buguk and Brorsen (2003) Turkey LoMacKinlay, Unit root, Wright VR, W. 1992.11999.12
Smith and Ryoo (2003) 18 European emerging ChowDenning MVR W, 1991:41998:8
countries
Lima and Tabak (2004) China, Hong Kong, Singapore ChowDenning MVR D, 1992:62000:12
Sheddighi and Nian (2004) China Unit root, ARCH D, 2000:12000:12
D: daily, W: weekly, M: monthly, Q: quarterly, Y: yearly.

References

Abraham, A., Seyyed, F. G., & Alsakran, S. (2002). Testing the random walk behaviour and efficiency of the Gulf stock markets. Financial Review,
37, 469480.
Al-Khazali, O. M., & Pyun, C. S. (2004). Stock prices and inflation: New evidence from the Pacific-Basin countries. Review of Quantitative Finance
and Accounting, 22, 123140.
Antoniou, A., & Ergul, N. (1997). Market efficiency, thin trading and non-linear behavior: Evidence from an emerging market. European Financial
Management, 3, 175190.
Ayadi, O. F., & Pyun, C. S. (1994). An application of variance ratio test to the Korean securities market. Journal of Banking & Finance, 18, 643658.
Buguk, C., & Brorsen, B. W. (2003). Testing weak-form market efficiency: Evidence from the Istanbul stock exchange. International Review of
Financial Analysis, 12, 579590.
Chang, E. J., Lima, E. J. A., & Tabak, B. M. (2003). Testing for weak form efficiency in emerging markets, Working Paper (Banco Central do Brasil,
Brasilia, Brazil).
Chang, K., & Ting, K. (2000). A variance ratio test of the random walk hypothesis for Taiwan's stock market. Applied Financial Economics, 10(5),
525532.
Chow, K. V., & Denning, K. C. (1993). A simple multiple variance ratio test. Journal of Econometrics, 58(3), 385401.
Darrat, A. F., & Zhong, M. (2000). On testing the random-walk hypothesis: A model comparison approach. Financial Review, 35, 105124.
Deo, R. S., & Richardson, M. (2003). On the asymptotic power of the variance ratio test. Econometric Theory, 19(2), 231239.
Efron, B., & Tibshirani, R. J. (1993). An introduction to the bootstrap. New York: Chapman and Hall.
El-Erian, M., & Kumar, M. (1995). Emerging equity markets in the Middle Eastern countries. IMF Staff Papers, 42, 313343.
Grieb, T., & Reyes, M. G. (1999). Random walk tests for Latin American equity indexes and individual firms. Journal of Financial Research, 22(4),
371383.
Hahm, J. H., & Mishkin, F. S. (2000). Cause of the Korean financial crisis: Lessons for policy, Working Paper 7483, National Bureau of Economic
Research, Cambridge, MA.
Hahn, G. J., & Hendrickson, R. W. (1971). A table of percentage points of the distribution of the largest value of K Student T variates and its
applications. Biometrika, 58, 323332.
Huang, B. N. (1995). Do Asian stock market prices follow random walks? Evidence from variance ratio test. Applied Financial Economics, 5,
251256.
Kawakatsu, H., & Morey, M. R. (1999). An empirical examination of financial liberalization and efficiency of emerging market stock prices. Journal
of Financial Research, 22, 358411.
Khil, J., & Lee, B. (2000). Are common stock a good hedge against inflation? Evidence from the Pacific-rim countries. Pacific-Basin Finance
Journal, 8, 457482.
502 H.A.A.B. Hoque et al. / International Review of Economics and Finance 16 (2007) 488502

Laurence, M., Cain, F., & Qian, S. (1997). Weak-form efficiency and Causality tests in Chinese stock markets. Multinational Finance Journal, 1(4),
291307.
Lee, C. F., Chen, G. M., & Rui, O. M. (2001). Stock returns and volatility on China's stock markets. Journal of Financial Research, 24(4), 523543.
Lima, E. J. A., & Tabak, B. M. (2004). Testing of the random walk hypothesis for equity markets: Evidence from China, Hong Kong and Singapore.
Applied Economics Letters, 11, 255258.
Lo, A. W., & MacKinlay, A. C. (1988). Stock market prices do not follow random walks: Evidence from a simple specification test. Review of
Financial Studies, 1(1), 4166.
Lo, A. W., & MacKinlay, A. C. (1989). The size and power of the variance ratio test in finite samples. Journal of Econometrics, 40(2), 203238.
Malkiel, B. G. (2003). The efficient market hypothesis and its critics. Journal of Economic Perspectives, 17, 5992.
Malliaropulos, D., & Priestley, R. (1999). Mean reversion in Southeast Asian stock markets. Journal of Empirical Finance, 6, 355384.
Nam, K., Pyun, C. S., & Kim, S. W. (2003). Is asymmetric mean-reverting pattern in stock returns systematic? Evidence from Pacific-basin markets in
the short-horizon. Journal of International Financial Markets, Institutions and Money, 13, 481502.
Perkins, D. H., & Woo, W. T. (1998). Malaysia in turmoil: Growth prospect and future competitiveness, Working Paper, Harvard University and
University of California at Davis.
Politis, D. N., Romano, J. P., & Wolf, M. (1997). Subsampling for heteroskedastic time series. Journal of Econometrics, 81, 281317.
Richardson, M., & Smith, T. (1991). Tests of financial models in the presence of overlapping observations. Review of Financial Studies, 4, 227254.
Ryoo, H. J., & Smith, G. (2002). Korean stock prices under price limits: Variance ratio tests of random walks. Applied Financial Economics, 12(7),
475484.
Savin, N. E. (1984). Multiple hypothesis testing. In Z. Griliches, & M. D. Intriligator (Eds.), Handbook of econometrics Oxford: North-Holland.
Seddighi, H. R., & Nian, W. (2004). The Chinese stock exchange market: Operations and efficiency. Applied Financial Economics, 14, 785797.
Slezak, S. L. (2003). On the impossibility of weak-form efficient markets. Journal of Financial and Quantitative Analysis, 38, 523554.
Smith, G., Jefferis, K., & Ryoo, H. (2002). African stock markets: Multiple variance ratio tests of random walks. Applied Financial Economics, 12(7),
475484.
Smith, G., & Ryoo, H. J. (2003). Variance ratio tests of the random walk hypothesis for European emerging stock markets. European Journal of
Finance, 9(3), 290300.
Stoline, M. R., & Ury, H. K. (1979). Tables of the studentized maximum modulus distribution and an application to multiple comparisons among
means. Technometrics, 21, 8793.
Urrutia, J. L. (1995). Tests of random walk and market efficiency for Latin American emerging markets. Journal of Financial Research, 18(3),
299309.
Van der Hart, J., Stagter, E., & van Dijk, D. (2003). Stock selection strategies in emerging markets. Journal of Empirical Finance, 10, 105132.
Whang, Y. J., & Kim, J. (2003). A multiple variance ratio test using subsampling. Economics Letters, 79, 225230.
Wright, J. H. (2000). Alternative variance-ratio tests using ranks and signs. Journal of Business & Economic Statistics, 18(1), 19.