Sei sulla pagina 1di 10

18

Asia-Pacific Business Review


Vol. V, No. 4, October - December 2009
pp. 18-27. ISSN: 0973-2470

An Empirical Analysis of Linear and Non-Linear


De
pendence of Stoc
k Prices and Tr ading Volume
Dependence
Stock
in Asia P
acif
ic Stoc
k Mar
ket
Pacif
acific
Stock
Mark
K. Srinivasan1, Malabika Deo2 and B. Murugesan3

ER
C

IA
L

U
SE

This paper examines the casual dependence of price changes and trading volume by using linear and nonlinear models for Australia, India, Japan, New Zealand and Taiwan stock exchange from January 1, 2005
to December 31, 2008. The empirical methods used include Unit-root tests, Granger causality tests and
MA (5) GARCH (1,1). Granger Causality test demonstrates that for some countries, returns cause volume
and volumes cause returns. The evidence indicates stronger evidence of returns causing volume than
volume causing returns. The results present a significant relationship between trading volume and
the value of price changes, the exogenous variable contributes some information to the return and
volatility series for capturing the potential non-linear dependence of stock and trading volume
in conditional variance to determine the contemporaneous and lagged volume effect after
incorporation. Moreover, the findings suggest that the presence of current and past returns,
trading volume adds some predictive power for future returns in these countries.

Introduction

the lead-lag relationship between stock return and


trading volume and referred as sequential information
arrival hypothesis (SEQ). It is really intricate; to test
the non-informational trade by using stock returns
data. The basic logic is to use trading activity
variable having explanatory power in addition to
past returns and price changes (Ali, 1997).

FO

C
O

The structural changes in trading activity are not


only explained by publicly available information but
also by non information trade due to events, short
selling, insider traders etc Campbell et al., (1993).
These factors are exogenous to the price behaviour
in stock market and considered to be critical
information which influences both future prices and
price volatility. On one hand, price movements and
trading volumes can be jointly considered as
aggregate market information. On the other hand,
trading volumes have the specificity of reflecting the
cumulative response of investors to news, whereas
price movements can only capture the impact of that
news on the average change in investors
expectation. Thus, trading volumes serves as a
critical component in exploring stock returns and
volatilities. The academic literature provides the
association between trading volume and stock return
volatility.

Keywords: Returns, Trading Volume, Mixture of Distribution, Granger Causality, GARCH

This paper adds to the growing literature by examining


the dependence of linear and non-linear relationship
between stock returns and trading volume data, and
makes two contributions. First, the positive
contemporaneous relationship between stock returns
and trading volume is called simultaneous information
arrival hypothesis (SIM). The second is to analyze

The rest of the paper is structured as follows. Section


2 provides an overview of the previous research on
the relation between stock prices, trading volume
and volatility. Section 3 describes the datasets
adopted for the study and the statistical tests and
models used in the analysis of data have been
discussed in Section 4. The results of Grangercausality tests and nonlinear dependence between
stock returns and trading volume in terms of GARCH
results and interpretations are carried out in Section
5. Finally, Section 6 summarizes and concludes.

Literature Review
There are number of empirical papers that provide
indirect evidence on the relationship between trading
volume and stock returns. The early literature was
first documented by Ying (1966). Similarly, positive
contemporaneous relationship between the variance
of price change and trading volume was linked by
Ragalski (1978), Figlewski and Cornell (1981). The

Department
of Commerce
(SOM),
Pondicherry Central University, Kalapet, Puducherry 605 014,Vol.
India V,
Asia-Pacific
Business
Review
1
E-mail: ksrinivasan1979@gmail.com, 2E-mail: deo_malavika@yahoo.co.in
3
E-mail: muru_gesan859@yahoo.co.in

No. 4, October - December 2009

Downloaded from abr.sagepub.com by SAGE Production (DO NOT CHANGE THE PASSWORD!) on January 6, 2016

19

four Asian stock markets using monthly data and


find strong evidence for causality running from
volume to absolute price changes and from price
changes to volume in all markets except Philippines.
Chorida and Swaminathan (2000) analysed the
correlation between volume and short-term returns
by concluding that trading volume plays a significant
role in propagating a wide range of market
information.

ER
C

IA
L

U
SE

Lamoureux and Lastrapes (1990) examined the


presence of ARCH/GARCH based on the hypothesis
that daily returns are generated by a mixture of
distributions using trading volume as a proxy for the
rate of daily information arrival, they find that
volatility persistence vanishes under the presence
of trading volume series in the conditional variance
equation. Martikainen et al., (1994) empirically
evidenced bidirectional feedback relation between
volume and stock prices in the period 198388. But
for the period from 197782, however, no causality
is observed. This significant variation in the results
over time is explained by the development of Finnish
financial market during the research period.
Several studies have followed Lamoureux and
Lastrapes (1990) to examine the relationship between
return volatility and information flow for other markets
Martikainen et al., (1994) and Pyun et al., (2000).
Omran and Mckenzie (2000) investigated the relation
between volume of trade and conditional variance of
trade and found significant relation between timing
of innovational outliers in returns and volume. Chen
et al., (2001) report that the persistence in volatility
is not eliminated when lagged or contemporaneous
trading volume level is incorporated into the GARCH
model, a result contradicting the findings of
Lamoureux and Lastrapes (1990). Brailsford (1996)
found the irrespective of direction in price change
was significant across three measures of daily trading
volume for the aggregate market and was significant
for individual stocks. Miyakoshi (2002) finds that
the inclusion of the trading volume variable in both
ARCH and exponential GARCH eliminate the ARCH/
GARCH effect for individual stocks listed on the
Tokyo stock Exchange and their price index.

linear and non-linear causality between the stock


prices and trading volume has also received a
substantial amount of attention in the literature
Hiemstra and Jones (1994). Campbell et al., (1993)
examined the relationship between aggregate stock
market trading volume and the serial correlation of
daily stock return. They found that a stock price
decline on high volume day is more likely than a
stock price decline on low volume day to be
associated with an increase in the expected stock
return. This investigation has also been extended to
bond and futures markets (Clark 1973; Hanna 1978;
Grammatikos and Saunders, 1986) and the examination
of cross-country spillovers between trading volume
and stock returns (Lee and Rui, 2002). In general,
both Mixture of Distribution Hypothesis and
sequential arrival of information hypothesis support
a positive and contemporaneous relationship between
volume and absolute return and assume symmetric
effects of price increase and decrease for futures
contracts (Karpoff, 1987). Pyun Lee and Nam (2000)
provide positive evidence from the Korean stock
market. Bohl and Henke (2003) show support for the
Polish stock market, while Luckey (2005) finds mixed
evidence for the mixture of distribution hypothesis
in the Irish stock market.

FO

Ragunathan and Pecker (1997) focus on the


relationship between volume and price variability for
the Australian futures market and explore positive
relationship between volume and volatility by
documenting asymmetric volatility response to
unexpected shocks in trading volume by using the
model developed by Bessembinder and Seguin (1993).
Positive unexpected shocks to trading volume were
found to induce an average increase in volatility at
76 per cent, while negative unexpected shocks to
trading volume induce a smaller response in volatility.
Daigler and Wiley (1999) examined the volumevolatility relation in futures markets for Chicago Board
of Trade for four types of traders.
Some theoretical studies explicitly investigate the
dynamic relation between trading volume and stock
returns, a relation that might have some causal relation
implications. Several studies have tested, by using
VAR, Granger Causality between the two series using
different samples and estimated techniques. Blume
et al., (1994) derives that investors can predict the
market information with past price and trading
volume. Wang (1994) shows that trade informational
and non-informational reasons will also lead to
different dynamic between trading volume and stock
returns. Moosa and Al-Loughani (1995) examined

In recent times numerous studies have been made at


the international level to test the causal nexus
between stock return and trading volume series,
whereas in India studies on the relationship between
these two variables are quite limited. Bhanupant
(2001) investigates the dynamic relationship between
stock return and trading volume using linear and
non linear Granger causality and evidenced that

Asia-Pacific Institute of Management, New Delhi

Downloaded from abr.sagepub.com by SAGE Production (DO NOT CHANGE THE PASSWORD!) on January 6, 2016

20

Methodology

the relationship have improved after rolling


settlement mechanism. Mahajan and Singh (2008)
examined the trading volume and return volatility
dynamics and reported weak unidirectional causality
from volume to return and indicate mild support for
sequential information from volume to price change.
Malabika et al., (2008) have tested the relationship
between these two variables for select Asia Pacific
stock market and that envisage bidirectional causality
exist between most of the stock exchanges. Hence,
the current study attempts to shed light on the
efficiency of simultaneous information arrival
hypothesis and sequential information arrival
hypothesis for select Asia Pacific stock market, to
fill the gap in the existing literature.

Augmented Dickey Fuller Test

U
SE

Augmented Dickey Fuller (ADF) test has implicitly


assumed that the estimated errors are statistically
independent and homoscedastic. Heteroskedasticity
does not affect a wide range of unit root test
statistics. However, a problem will occur if the
estimated residual t is not free from autocorrelation
since this invalidates the test. If the error term t is
auto correlated. The Augmented Dickey Fuller test is
given by the following equations;

Data Source

ER
C

IA
L

where, the lagged difference terms are being


determined by minimum number of residuals free
from autocorrelation. This could be tested for in the
standard way such as Akaike Information Criterion
(AIC) or Schwartz Bayesian Criterion (SIC). The
statistics are all used to test hypothesis = 0, i.e.,
there exists a unit root.

Granger Causality Test

Granger Causality (1969) test was employed to


investigate the causal nexus between two variables,
says yt and xt affect each other with distributed lags.
The relationship between those variables can be
captured by VAR model. In this case, it is possible
to have that yt causes xt or xt causes yt, then there
exists a bi-directional feedback, and finally the two
variables are independent. The problem is finding an
appropriate procedure that allows us to test and
statistically detect the cause and effect relationship
among the variables. The Granger test can detect the
direction of causality when temporarily there is a
lead lag relationship between two variables. Many
authors have used the Granger causality to analyze
the lead lag relationship between stock returns
and trading volume Chen and Firth (2001) Malabika
and Srinivasan (2008).

FO

The data set comprises of the bivariate stock market


data in terms of returns and trading volume series
for the five largest Stock Exchanges in the Asia
Pacific region: Australia, India, Japan, New Zealand
and Taiwan. We use All Ordinary Index (AORD) for
Australia, Bombay Stock Exchange (BSE) for India,
(Nikkei, 225) for Japan, New Zealand Stock Exchange
(NZSE 50) for New Zealand and Taiwan Composite
(TW 11) for Taiwan. For the stock index, we first
choose 11 stock exchanges according to their market
capitalization and then we selected those indexes
whose data were available since 2005:01:01. Only
five Stock Exchanges meet this criterion and we
choose these markets. The data sets spreads from
2005:01:01 to 2008:12:31 yielding more than 1000
observations for Australia and less than 1000
observations for other stock markets. The sample of
data used in this exercise is the daily opening, high,
low closing prices and trading volume for select
Asia Pacific stock market. In this study each days
index value is calculated by taking the average of
these four variables. This procedure is adopted to
avoid the assumption that trading is done at the
closing value Sarma (2004). Daily closing prices and
trading volumes are retrieved from DataStream. An
adjusted return was calculated as Rt = log (Pt/Pt-1)
where Pt and Pt-1 are natural logarithms of adjusted
return on day t and t-1 respectively. The logarithm
of price relative is used to calculate the price change.
The use of logarithmic price changes prevents nonstationarity of the price level data affecting future
price variability. As far as the trading volume is
concerned, the study applies logarithmic procedure
on other variables to account for the non
stationarity in the series.

Asia-Pacific Business Review

Once the test for unit roots has been conducted, we


then estimate the VAR model by the following
equation.

Vol. V, No. 4, October - December 2009

Downloaded from abr.sagepub.com by SAGE Production (DO NOT CHANGE THE PASSWORD!) on January 6, 2016

21

where, Rt and Vt are stock returns and trading volume.


The optimal lag structure for VAR model for each
and every market was categorized on several criteria,
which incorporate measures of the precision and
parsimony in the parameterization of the models in
hand Akaike Information Criterion (AIC) and
Schwatzs Bayseian Information Criterion (SIC) and
Hannan and Quin Criterion (HQC).

In order to estimate the volatility persistence between


the bivariate variable, we included lagged return and
trading volume in the GARCH model specification.
Engle (1982) and Bollerslev (1986) emphasized the
inclusion of exogenous variables in the conditional
variance; the volume series are included to evaluate
their incremental significance in return prediction
(Bollerslev et al., 1992). The model to be estimated
for each index in the sample is given by the
generalized variance equation specification

U
SE

where, SSEr stands for the sum of squared residuals


of the restricted regression, SSEf is the sum of squared
residuals of the unrestricted equation. Finally T and
m refer to number of observations and their optimal
lag structure respectively.

ER
C

Results and Interpretation

FO

To investigate the linear and non linear dependence


between stock returns and trading volumes GrangerCausality tests and GARCH models are adopted.
The first ARCH model was presented by Engle (1982).
The model suggests that the variance of the
residuals at time t depends on the squared error
terms from past periods. The residual term it is
conditionally normally distributed and serially
uncorrelated. Bollerslev (1986) generalized the ARCH
models by allowing the conditional variance, h, to be
dependent on the last periods conditional variance.
The conditional variance in GARCH (1,1) model can
be defined as follows;

Under the assumption that volume (3) is the mixing


variable, in the sense of Engle, Lilien and Robins
(1987) volume series is considered as a exogenous
variable. We restrict our focus to GARCH (1,1)
specification because it is considered to be the
parsimonious representation of conditional variance
that adequately fits many econometric time series
models.

IA
L

GARCH Modelling

Numerous studies of recent origin reported that one


lagged conditional variance term appears to model
conditional variance adequately in US stock market
(Bollerslev, 1987; Akgiray, 1989). To capture the
autocorrelation of conditional mean return, the
adjusted nth order moving average for MA (1)
GARCH (1,1) model can be expressed as follows;

The basic statistical characteristics for stock return


and trading volume series are presented in Table 1.
Out of five indices, the highest observation was
observed for AORD and lowest for Nikkei 225. AORD
have negative mean returns, and the rest of the
indices have positive trading volume and mean
return. In stock return and trading volume the market
volatility was observed higher for Nikkei 225 and
AORD with 0.0652 and 0.4464 respectively. In
skewness, we observed the right tail was particularly
extreme for Nikkei 225, NZ 50 and TW 11 stock
returns. The kurtosis value for selected Asia Pacific
stock returns signifies with leptokurtic, but the
trading volume series identifies with less than three
for Nikkei 225 and TW 11. Significant deviations
from normality can be symptom of non linear
dependence. As a result, the Jarque Bera test
rejects the null of the normality. In order to check
the stationarity of the return and trading volume
series, we used Augmented Dickey Fuller (ADF) test.
The test results show that the bivariate variables
exhibits stationary at its level.

Asia-Pacific Institute of Management, New Delhi

Downloaded from abr.sagepub.com by SAGE Production (DO NOT CHANGE THE PASSWORD!) on January 6, 2016

22

Table 1: Summary Statistics of Stock Return & Trading Volume for Select Asia Pacific
Stock Exchange

TW11 AORD
967
1014
0.0045 20.3977
0.0616 0.4464
0.9790 -4.8566
4.5500 49.5806
251.30 9480.49
0.0000 0.0000
3397.8 386.44
5273
557.94
-4.289* -7.691*

Trading Volume
Nikkei
BSE
225
NZ 50 TW11
988
964
970
967
9.8336 11.6965 17.1319 15.1465
0.4371
0.2922
0.3752 0.3038
0.5413 -0.0215
0.2423 0.2338
5.4494
2.9045
5.5284 2.8986
294.645 0.4477 271.745 9.1344
0.0000
0.0000
0.0000 0.0000
1535.9
2280.1
236.71 2867.8
2477
3772
250.52 4797.9
-3.390* -4.910* -9.435* -3.821*

U
SE

Particulars AORD
NOB
1014
Mean
-0.0001
Std. Dev.
0.0126
Skewness
-0.6508
Kurtosis
9.2365
Jarque-Bera 1714.89
Probability
0.0000
LB[5]
15.647
LB[10]
26.953
ADF
-12.972*

Stock Index Returns


Nikkei
BSE
225 NZ 50
988
964
970
0.0003
0.0050 0.0023
0.0190
0.0652 0.0384
-0.3989 2.0547 0.7900
6.3089 11.2158 4.3311
476.94 3389.54 172.53
0.0000
0.0000 0.0000
9.4719
3269.8 3592.8
19.726
4834.4 5660.5
-22.296* -4.227* -4.241*

IA
L

Note: LB [5] and [10] represents the Ljung Box Q Statistics for Autocorrelation * indicates Significant at 1 per cent
level.

FO

Table 2 presents the results of causal relation test


based on a bivariate model. F-Statistics and their
corresponding significance levels are also shown in
Panel A and Panel B. Panel A reveals the results to
test the null hypothesis that returns do not cause
volume. The F-Statistics is significant at 1 per cent
level for BSE, Nikkei 225, NZ 50 and TW 11. In Panel
B, shows the test of the null hypothesis, volume
does not cause return. The F-Statistics are

statistically significant at 5 per cent level for TW 11


and significant at 1 per cent level for BSE and Nikkei
225. The findings suggest that the presence of
current and past returns, trading volume adds some
predictive power for future returns in these countries.
The results from Panel A and Panel B imply a
feedback system was prevailing in BSE, Nikkei 225
and TW 11. In these three countries, returns are
influenced by volume and volume is influenced by
returns.

ER
C

Results from Granger Causality Tests

Table 2: Granger Causality Test

AORD
BSE
0.186* [5.99]
0.078** [2.44]
0.049 [1.57]
-0.051 [-1.60]
0.092* [2.94]
-0.037 [-1.17]
0.080* [2.55]
-0.037 [-1.17]
0.216* [7.06]
-0.022 [-0.71]
-4.615* [-4.65]
-0.001 [-0.66]
-1.655 [-1.65]
0.000 [0.23]
-2.138** [-2.13]
0.001 [0.59]
-3.770* [-3.77] -0.005* [-2.68]
-0.491 [-0.48]
0.000 [0.44]
7.680* [8.47]
0.042** [2.50]
0.222
0.583*
28.355
2.311

Coefficient
Rt-1
Rt-2
Rt-3
Rt-4
Rt-5
Vt-1
Vt-2
Vt-3
Vt-4
Vt-5
c
R-Squared
F-Statistic

Panel A: Estimation Results between Stock Return and Trading Volume


Nikkei 225
NZ 50
0.523* [16.23]
1.039* [32.05]
0.148* [4.06]
-0.180*[-3.85]
0.021 [0.59]
0.152* [3.24]
0.070** [1.92] -0.086** [-1.83]
0.125* [3.91]
0.026 [0.79]
-0.546** [-2.33]
0.001 [1.40]
0.437 [1.34]
-0.000 [-0.20]
0.240 [0.73]
-0.000 [-0.51]
-0.046 [-0.14]
0.001 [1.33]
0.055 [0.23]
0.000 [0.50]
1.288* [5.17]
-0.048** [-1.71]
0.669*
0.907*
191.868
935.257

TW11
0.962* [29.63]
0.105** [2.33]
-0.185* [-4.11]
0.009 [0.19]
0.047 [1.46]
-0.008** [-2.04]
0.013 [2.78]
-0.003 [-0.71]
0.002 [0.45]
-0.009** [-2.10]
0.079** [2.17]
0.899*
846.577

Note: *Denotes Significance at 99% Confidence Interval and ** denotes Significance at 95 % Confidence Interval.

Asia-Pacific Business Review

Vol. V, No. 4, October - December 2009

Downloaded from abr.sagepub.com by SAGE Production (DO NOT CHANGE THE PASSWORD!) on January 6, 2016

23

Panel B: Estimation Results between Trading Volume and Stock Return


AORD
BSE
-0.001[-1.182] 1.146** [-2.21]
0.001[1.128]
-0.330 [-0.637]
0.000 [0.402]
-0.680 [-1.311]
0.000 [0.242]
0.331 [0.639]
-0.002 [-1.528] -0.279 [-0.540]
-0.042 [-1.322] 0.341* [10.71]
-0.015 [-0.453] 0.182* [5.441]
-0.075* [-2.841] 0.082** [2.431]
0.0669* [2.986] 0.117* [3.502]
0.0303 [0.939]
0.133* [4.187]
0.018 [0.625]
1.402* [5.136]
0.018
0.525*
1.9144
107.4839

Nikkei 225
-0.000 [-0.02]
-0.004 [-0.88]
0.006 [1.35]
-0.004 [-0.91]
0.002 [0.64]
0.962* [29.75]
-0.073 [-1.63]
0.028 [0.63]
0.121* [2.69]
-0.107* [-3.28]
-0.006 [-0.17]
0.875*
664.5357

NZ 50
-2.216* [-2.762]
1.729 [1.278]
-1.421 [-1.048]
1.713 [1.264]
0.065 [0.069]
0.318* [9.942]
0.048 [ 1.450]
-0.015 [-0.469]
0.066** [1.973]
0.138* [4.331]
7.603* [9.304]
0.169
19.3645

U
SE

Coefficient
Vt-1
Vt-2
Vt-3
Vt-4
Vt-5
Rt-1
Rt-2
Rt-3
Rt-4
Rt-5
c
R-Squared
F-Statistic

TW11
0.151 [0.615]
-0.084 [-0.246]
0.113 [0.332]
-0.489 [-1.414]
0.221 [0.890]
0.601* [18.52]
0.136* [3.583]
0.0371 [0.964]
0.063** [1.656]
0.081* [2.473]
1.227* [4.449]
0.762*
301.7720

ER
C

IA
L

Note: *Denotes Significance at 99% Confidence Interval and ** denotes Significance at 95 % Confidence Interval.

Results from GARCH Models

Return Prediction

FO

Table 3 reports the results from the basic modelling


of return series. Regarding returns, MA (5) GARCH
(1,1) models were found to be appropriate optimum
lag structure using the Akaikes final prediction error
(FPE) criterion. We obtain the parameter estimates
by maximizing the log likelihood using the (Berndt,
Hall, Hall and Hausman-1974) algorithm.

In Panel A, the coefficient estimates of MA (5) model


are found to be insignificant for AORD and BSE and
the GARCH (1,1) model performed best for all the
stock indices. The F Statistics are significant at 5
per cent level for TW 11 and significant at 1 per cent
level for Nikkei 225 and NZ 50. The log likelihood
statistics are very high for Nikkei 225, NZ 50 and TW
11 stock indices and this allow us to test the null
hypothesis of the return series are normally distributed

against the alternative that they are generated by


MA (5) GARCH (1,1) model. This results shows that
the existence of changes in conditional variances in
return series in the Asia pacific stock exchanges.
Panel B in Table 3 introduces an exogenous variable
into the conditional variance to capture the potential
non linear dependence of stock return series. Since
the residuals from Panel A can be interpreted as an
unpredictable component for the returns, the most
recent residuals series is included in the conditional
variance of stock return specifications. More
precisely, the estimated parameter 3 indicates with
statistically significant for all the stock indices except
New Zealand. The F Statistics is significant at 1
per cent level for BSE, Nikkei 225 and TW 11. This
result is consistent with the earlier findings of
Lamoreux and Lastrapes (1991), Martikainen et al.,
(1994)

Asia-Pacific Institute of Management, New Delhi

Downloaded from abr.sagepub.com by SAGE Production (DO NOT CHANGE THE PASSWORD!) on January 6, 2016

24

Table: 3 Linear and Non - Linear Models for Stock Return Series
Panel: A Estimation Results of MA (5) GARCH (1,1) Model for Stock Return
Nikkei 225
-0.011* [-4.82]
0.967* [27.08]
0.875* [19.93]
0.754* [15.69]
0.574* [13.63]
0.335* [9.95]
0.000* [2.15]
0.130* [6.50]
0.870* [44.39]
621.9909*
2431.906

NZ 50
-0.007* [-4.16]
1.078* [33.95]
0.955* [24.10]
0.858* [20.39]
0.765* [19.18]
0.441* [14.79]
0.000* [2.449]
0.127* [5.57]
0.856* [32.45]
825.8190*
2960.545

U
SE

Coefficient
AORD
BSE
C
0.000* [3.40]
0.001* [4.06]
MA[1]
-0.075** [-2.05] 0.084** [2.27]
MA[2]
0.002 [0.07]
-0.045 [-1.31]
MA[3]
0.048 [1.49]
0.017 [0.52]
MA[4]
0.004 [0.13]
-0.017 [-0.54]
MA[5]
-0.054 [-1.58] -0.057** [-1.67]

0.000* [3.26]
0.000* [4.31]

0.151* [8.01]
0.150* [7.61]

0.841* [40.32]
0.829* [40.96]
F - Statistics
28.36
7.33E-05
Log Likelihood
3242.938
2714.190

TW11
-0.012* [-4.72]
1.005* [33.43]
0.972* [26.64]
0.848* [19.83]
0.691* [18.21]
0.412* [13.56]
0.000** [1.92]
0.087* [4.93]
0.909* [52.14]
787.5731*
2463.707

ER
C

IA
L

Note: * Denotes Significance at 99% Confidence Interval and ** denotes Significance at 95 % Confidence Interval.

Panel: B Estimation Results of MA (5) GARCH (1,1) Exogenous Variables in Stock Return
BSE
-0.000 [-0.82]

MA[1]

-0.062* [-5.42]

0.127 [6.89]

MA[2]

0.019** [1.65]

-0.007 [-0.43]

MA[3]

0.070** [6.94]

0.051* [3.21]

Nikkei 225
-0.012* [-34.11]

NZ 50
-0.006* [-5.04]

TW11
-0.003 [-1.22]

0.949* [17.47]

1.082* [49.01]

0.983* [45.66]

0.849* [98.74]

0.968* [34.67]

0.986* [38.35]

0.732* [10.16]

0.864* [39.34]

0.853* [33.89]

AORD
0.000 [0.02]

0.025* [2.11]

0.007 [0.48]

0.553* [76.78]

0.767* [35.12]

0.784* [32.25]

MA[5]

-0.044* [-4.79]

-0.031** [-1.98]

0.447* [24.30]

0.460* [20.44]

0.000* [7.70]

0.325* [50.67]

0.000* [6.82]

0.000* [2.55]

0.000* [2.63]

0.001* [30.02]

-1.486* [-3.00]

-0.727* [-12.04]

1.589* [83.22]

-0.512 [-0.85]

0.084* [4.23]

0.836* [33.01]

0.839* [48.89]

0.897* [49.62]

0.856* [30.60]

-0.998* -[29.03]

1.599* [3.19]

0.844* [13.76]

-1.496* [-15.77]

0.630 [1.05]

-0.060* [-3.29]

F - Statistics

153.05

88.37*

547.8337*

737.3593*

715.9409*

3274.560

2733.324

2439.934

2958.020

2314.309

Log Likelihood

FO

MA[4]

Coefficient
C

Note: * Denotes Significance at 99% Confidence Interval and ** denotes Significance at 95 % Confidence Interval.

Volume Prediction
Table 4 in Panel A and Panel B reveals the linear and
non-linear dependence of trading volume series and
introduces an exogenous variable into the conditional
variance of trading volume included in Panel B. The
F Statistics ate statistically significant at 5 per cent
level for AORD and NZ 50 with 8.365 and 21.724
respectively. The log likelihood statistics in Panel
A are highly significant. This supports the existence

Asia-Pacific Business Review

of conditional variance in trading volume series in


the Asia Pacific stock indices. In Panel B an
exogenous variable was introduced again into the
conditional variance. This variable captures the
potential non linear dependence from yesterdays
stock return into the volume. Moreover, the parameter
estimates 3 on the volume with significant for AORD,
BSE and NZ 50. The F Statistics envisages with
statistically significant for all the Asia Pacific stock
market at 1 per cent level.

Vol. V, No. 4, October - December 2009

Downloaded from abr.sagepub.com by SAGE Production (DO NOT CHANGE THE PASSWORD!) on January 6, 2016

25

Table: 4 Linear and Non - Linear Models for Trading Volume Series
Panel: A Estimation Results of MA (5) GARCH (1,1) Model for Trading Volume
Nikkei 225
11.708* [51.79]
0.608* [17.71]
0.516* [12.71]
0.371* [9.35]
0.269* [6.96]
0.187* [5.44]
0.002* [2.51]
0.072* [3.47]
0.846* [17.67]
181.4629*
284.5190

NZ 50
17.157* [79.14]
0.321* [8.82]
0.163* [5.04]
0.029 [0.85]
0.085* [2.85]
0.187* [5.99]
0.110* [14.69]
0.190* [4.79]
-0.120* [-3.42]
21.72437*
-320.4170

U
SE

BSE
9.816* [37.82]
0.545* [14.78]
0.471* [13.23]
0.365* [9.51]
0.236* [5.88]
0.169* [5.06]
0.023* [8.10]
0.246* [8.74]
0.56* [15.54]
95.27715*
-234.4007

TW11
15.150* [10.73]
0.693* [20.77]
0.551* [13.51]
0.444* [10.63]
0.408* [10.53]
0.278* [8.62]
0.002 [1.42]
0.060* [2.26]
0.834* [8.76]
279.8364*
374.1617

IA
L

Coefficient
AORD
C
20.307* [12.35]
MA[1]
0.483* [19.29]
MA[2]
0.281* [13.44]
MA[3]
0.136* [3.68]
MA[4]
0.228* [7.43]
MA[5]
0.152* [8.05]

0.023* [9.27]

1.533* [17.88]

0.272* [8.87]
F - Statistics
8.365203*
Log Likelihood
-354.1870

ER
C

Note: * Denotes Significance at 99% Confidence Interval and ** denotes Significance at 95 % Confidence Interval.

Panel: B Estimation Results of MA (5) GARCH (1,1) Exogenous Variables in Trading Volume
20.432* [13.09]
0.425* [19.11]
0.267* [15.03]
0.264* [10.26]
0.273* [12.18]
0.235* [19.35]
0.024* [12.82]
-0.892* [-5.71]
0.279* [17.21]
2.119* [15.07]
13.11545*
-312.7010

9.854* [53.11]
0.554* [38.57]
0.408* [15.01]
0.318* [11.81]
0.305* [11.59]
0.182* [9.11]
0.033* [7.80]
-0.881* [-3.07]
0.468* [8.15]
1.086* [3.68]
84.71244*
-219.0216

FO

Nikkei 225

NZ 50

TW11

11.689* [80.61]
0.638* [21.14]
0.523* [17.88]
0.398* [13.44]
0.284* [11.57]
0.182* [9.66]
0.020* [4.52]
0.710 [1.54]
0.262* [1.83]
-0.574 [-1.25]
160.7329*
280.8368

17.136* [17.20]
0.297* [15.59]
0.146* [7.79]
0.083* [4.44]
0.062* [2.97]
0.154* [8.23]
0.086* [6.47]
-1.181* [-3.82]
0.067 [0.57]
1.359* [4.30]
19.15131*
-308.7779

15.151* [43.07]
0.699* [77.03]
0.544* [43.66]
0.419* [17.38]
0.393* [24.05]
0.281* [38.06]
0.002* [2.29]
1.461 [1.06]
0.859* [14.78]
-1.423 [-1.03]
247.4544*
380.5709

BSE

C
MA[1]
MA[2]
MA[3]
MA[4]
MA[5]

F - Statistics
Log Likelihood

AORD

Coefficient

Note: * Denotes Significance at 99% Confidence Interval and ** denotes Significance at 95 % Confidence Interval.

Summary and Conclusion


This paper documents the empirical relationship
between return and trading volume series for
Australia, India, Japan, New Zealand and Taiwan
stock exchange during the period from 1st January
2005 to 31st December 2008. Before carrying out the
causal relationship, it is necessary to test the
stationarity of the series. Our study analysed the
dependence of return and trading volume relationship
for Asia Pacific stock market. In addition, our

investigations cover not only contemporaneous but


also dynamic relationships.
First, we test the bidirectional relationship between
trading volume and stock return series. Granger
Causality tests were employed to identify the
appropriate representation of the returns series
causing volume and volume series cause returns.
Our results show a positive correlation between
return and trading volume for India, Japan and Taiwan
stock exchanges, apart from this the return series are

Asia-Pacific Institute of Management, New Delhi

Downloaded from abr.sagepub.com by SAGE Production (DO NOT CHANGE THE PASSWORD!) on January 6, 2016

26

influencing trading volume but on the other side


trading volume series are not causing return for
Australia and New Zealand stock markets.

Bhanupant (2001), Testing Dynamic Relationship


between Returns and Trading Volume on the National
Stock Exchange, NSE Working Paper 2001,
www.utiicm.com/bhanupant.html.

Secondly, our results indicate that there is a strong


association in either direction between returns and
trading volume for all the stock markets, except
Australia in return series. This implies that the
knowledge of trading volume cannot improve shortrun return forecasts. On the other hand, we find
strong support for the hypothesis favouring that
there is a positive relationship between return
volatility and trading volume series as indicated by
the studies. (Karpoff, 1987; Lamoureux and Lastrapes,
1990).

Bollerslev, T. (1986), Generalized Autoregressive


Conditional Heteroskedasticity, Journal of Econometrics,
Vol. 31, pp. 307-327.

U
SE

Bollerslev, T. (1987), A Conditionally Heteroskedasticity


Time Series model for Speculative Prices and rates of
return, Review of Economics and Statistics, Vol. 69, pp.
116-131.

IA
L

Bollerslev, T., Chou, R. Y. and Kroner, K. F. (1992),


ARCH Modelling in Finance: A Review of the Theory
and Empirical Evidence, Journal of Econometrics, Vol.
52 (1&2), pp. 5-59.

ER
C

Bohl, M. T. and Henke, H. (2003), Trading Volume and


Stock Market Volatility: The Polish Case, International
Review of Financial Analysis, Vol. 12, pp. 513-525.

Brailsford, T. J. (1996), The Empirical Relationship


Between Trading Volume, Returns and Volatility,
Accounting and Finance, Vol. 36, pp. 89-111.

FO

Third, the linear and non-linear relationship is used


to capture the inter relationship between the return
and trading volume series with the help of residual
variables, and this suggest non linear model
performed best for all the stock exchanges. Our
evidence indicates, that the presence of current and
past returns, trading volume adds some predictive
power for future returns in these countries.
Moreover, the global factors continue to have an
effect on the stock market, particularly from US
stock movements; the strength of the local
economy underpins and stabilizes many sector
valuations. Finally, the deviation of residual series
is mainly attributing the return and trading volume
series and shows a positive strength across the Asia
Pacific stock markets and reveals that volume figures
can provide important information for both
practioners and researchers.

Blume, L., D. Easley and M. OHara (1994), Market


Statistics and Technical Analysis: The role of Volume,
Journal of Finance, Vol. 49 (1), pp. 153-182.

References

Akgiray, V. (1989), Conditional Heteroskedasticity in


Time Series of Stock Returns: Evidence and Forecasts,
Journal of Business, Vol. 62, pp. 55-80.
Ali, S. S. (1997), Prices and Trading Volume in Pakistan
Stock Markets, Journal of Economic
Cooperation among Islamic Countries, Vol. 18 (3), pp.
115-137.
Berndt, E. K., Hall, B. H., Hall, R. E. and Hausman, J.
A. (1974), Estimation and Inference in Nonlinear
Structural Models, Annals of Economic and Social
Measurement, Vol. 4, pp.653-665.
Bessembinder, H. and Seguin, P. (1993), Price Volatility,
Trading Volume, and Market Depth: Evidence from Futures
Markets, Journal of Financial and Quantitative Analysis,
Vol. 29, pp. 21- 39.

Asia-Pacific Business Review

Campbell, J., S. Grossman and J. Wang (1993), Trading


Volume and Serial Correlation in stock Returns, Quarterly
Journal of Economics, Vol. 108, pp. 905-939.

Chen, G., M. Firth and O.M. Rui, (2001), The Dynamic


Relation between Stock Returns, Trading Volume, and
Volatility, The Financial Review, Vol. 38, pp. 153-174.
Chorida, T. and Swaminathan. B (2000), Trading Volume
and Cross-Autocorrelations in Stock Returns, Journal of
Finance, Vol. 55, pp. 913-935.
Clark, P. (1973), A Subordinated Stochastic process model
with Finite Variances for Speculative Prices, Econometrica,
Vol. 41, pp. 135-155.
Daigler, R. and Wiley. M. (1999), The Impact of Trader
Type on the Futures Volatility-Volume Relation, Journal
of Finance, Vol. 54, pp. 2297-2316.
Engle, R. F. (1982), Autoregressive conditional
Heteroskedasticity with estimates of the variance of U.K.
Inflation, Econometrica, Vol. 50, pp. 987-1008.
Engle, R. F., Lilien, D. M. and Robins, R. P. (1987),
Estimating Time-Varying Risk Premia in the term
Structure: The ARCH-M Model, Econometrica, Vol. 55,
pp. 391-407.

Vol. V, No. 4, October - December 2009

Downloaded from abr.sagepub.com by SAGE Production (DO NOT CHANGE THE PASSWORD!) on January 6, 2016

27

Figlewski, S. Cornell (1981), Futures Trading and


Volatility in the GNMA Market, Journal of Finance,
Vol. 36, pp. 445-456.

Malabika. D, Srinivasan. K and Devanadhen. K (2008),


The Empirical Relationship between Stock Returns,
Trading Volume and Volatility: Evidence from Select AsiaPacific Stock Market European Journal of Economics,
Finance and Administrative Sciences, Vol. 12, pp. 58-68.

Grammatikos, T. and A. Saunders (1986), Futures Price


Variability: A test of Maturity and Volume Effects,
Journal of Business, Vol. 59, pp. 319-330.

Martikainen. T, Puttonen V. and Luoma M. (1994), The


Linear and Non-Linear dependence of Stock Returns and
Trading Volume in the Finnish Stock Market Applied
Financial Economics, Vol. 4, pp. 159-169.

Granger, C. W. J., (1969), Investigating Causal Relations


by Econometric Models and Cross Spectral Methods,
Econometrica, Vol. 37, pp. 424-438.

Miyakoshi, T. (2002), ARCH versus Information-Based


Variances: Evidence from the Tokyo Stock Market, Japan
and the World Economy, Vol. 14 (2), pp. 215-231.

Hanna, M. (1978), Security Price Changes and


Transaction Volumes: Additional evidence, American
Economic Review, Vol. 68, pp. 692-695.

Moosa, I. A., and Al-Loughani, N.E. (1995), Testing the


Price-Volume relation in emerging Asian Stock Markets,
Journal of Asian Economics, Vol. 6, pp. 407-422.

U
SE

Hiemstra, C. and J. D. Jones (1994), Testing for linear


and non-linear Granger causality in the Stock Price-Volume
Relation, Journal of Finance, Vol. 49, pp. 1639-1664.

Omran,
M.F.
and
McKenzie,
E.
(2000)
Heteroskedasticity in stock returns data revisited: Volume
versus GARCH effects, Applied Financial Economics,
Vol. 10, pp. 553-560.

ER
C

Pyun, C. S., Lee, S. Y. and Nam, K. (2000), Volatility


and information flows in emerging equity market A Case
of the Korean Stock Exchange, International Review of
Financial Analysis, Vol. 9, pp. 405-420.

Lamoureux, C. G. and Lastrapes, W. D. (1990),


Heteroskedasticity in stock return data: volume versus
GARCH effects, Journal of Finance, Vol. 45, pp. 221229.

IA
L

Karpoff, J. M. (1987), The relations between Price


Changes and Trading Volume: A survey, Journal of
Financial and Quantitative Analysis, Vol. 22, pp. 109126.

Lee, B. S. and Rui, O. M. (2002), The Dynamic


Relationship between Stock Returns and Trading Volume:
Domestic and Cross-Country Evidence, Journal of
Banking and Finance, Vol. 26, pp. 51-78.

FO

Luckey, B. M., (2005), Does Volume provide


Information? Evidence from the Irish Stock Market,
Applied Financial Economics Letters, Vol. 1, pp. 105-109.

Sarma, S. N. (2004), Stock, Market Seasonality in an


Emerging Market, Vikapla, Vol. 29 (3), pp. 35-42
Ying, C. C. (1966), Stock Market Prices and Volumes of
Sales, Econometrica, Vol. 34, pp. 676-686.
Wang, J. (1994), A model of Competitive Trading
Volume, Journal of Political Economy, Vol. 102, pp. 127168.

Mahajan. S. and Singh B. (2008), Trading Volume and


Return Volatility Dynamics in Indian Stock Market, ICFAI
Journal of Applied Finance, Vol. 14 (2), pp. 53-73.

Ragunathan. V and A. Peker (1997), Price Variability,


Trading Volume and Market Depth: evidence from the
Australian Futures Market, Applied Financial Economics,
Vol. 7, pp. 447-454.

Asia-Pacific Institute of Management, New Delhi

Downloaded from abr.sagepub.com by SAGE Production (DO NOT CHANGE THE PASSWORD!) on January 6, 2016

Potrebbero piacerti anche