ABSTRACT
 
Past studies in the United States, Canada and
Australia indicated the existence of seasonality
effect, i.e. "January effect" and "end-of-the-yea
effect " due to taxes imposed on the capital gains
in these countries.  This paper attempted to find
out whether or not this conclusion is also true in
the case of Malaysia where there is no tax imposed
or the capital gains. Overall, initial observation
showed that the average returns are higher in
January compared to other months for 5 out of 6
sectors of the Kuala Lumpur Stock Exchange.
However, the F-statistic test indicated that the
overall differences among these months were not'
statistically significant, which means there is no
seasonality effect in Malaysia. The t-statistic test
which made a detailed comparison between the average
returns in January with that of other individual
month found that, except for a few cases, the
average returns in January were not that much
different from other months.
 
 
INTRODUCTION
 
Numerous studies. in the West indicated some
anomalies in the validity of the efficiency of the
stock markets. Among them are the size effect,
Monday or weekend effect, year-end effect and the
January effect. It is found that the average returns
in January are higher than any other month and this
phenomenon is called the January effect. This
situation was attributed to the action by the
investors in response to taxes on the capital gains.
That is, the investors sell the securities with
losses at the end of the year to take advantage of
the tax rebate and delay the tax payments on the
capital gains by holding the securities with capital
gains. This action will result in the decrease of
securities' prices at the end of the year, and on
the other hand the increase in the prices of
securities in the month of January.  Keim (1983)
studied the securities traded on the New York Stock
Exchange (NYSE) for the period of 1963 to 1979 and
concluded that nearly 50 percent of these stocks
exhibited high returns in the month of January. He
also found that over 50 percent of those returns
occurred during the first week of January.
 
Rozeff and Kinney (1976) who studied the average
monthly returns on the NYSE between 1904 and 1974
found that the average return in January was higher
than any other month, except from 1929 to 1940.
They also found that the average returns for the
months of July, November, and December were quite
high and the lowest average returns were in the
months of February and June.
 
Dyl (1977) studied 100 firms selected randomly from
1948 to 1970 to determine whether the January effect
could be attributed to the taxes on capital gains.
He found high volumes of trading for the securities
with losses and low volumes of trading for
securities with capital gains. He concluded that
taxes on capital gains were the reason for the
January effect.
 
Givoly and Ovadia (1983) also found that the high
returns in January were due to the tax factor. Their
study which was done for the period of 1945 to 1979
found that the tax factor made the security prices
lower in December and higher in January.
 
Gultekin and Gultekin (1983) found that the average
rates of returns were unusually high in January for
nearly every country represented, as shown in
Appendix I. In addition, a study by Tinic and West
(1984) found that January is the only month of the
year where there exists a positive trade-off between
the beta of a stock and its realized rates of
return.
 
A study in Canada by Berges, McConnell and
Schlarbaum (1984) for the 391 firms traded on the
Toronto Stock Exchange from 1950 to 1980 found the
existence of the year-end effect and the January
effect as in the United States. They also found that
the effects were greater for small sized firms, thus
conforming the findings of the studies by Reinganum
(1981) and Roll (1982).
 
A study by Brown, Keim, Keildon and Marsh (1983)
from 1958 to 1979 of all industrial firms traded on
the Australian Stock Exchange found high average
returns for December - January and July - August.
These findings were similar with the findings of
Praetz (1973) where the average returns were high in
January - February and July - August but low average
returns in March - April and November- December.
 
The purpose of this paper is to find out whether the
seasonal effect exists in all six sectors of the
Kuala Lumpur Stock Exchange (KLSE) eventhough there
are no taxes on capital gains in Malaysia. This
paper will also identify whether or not there exists
a certain month which provides the highest return as
found in the United States, Australia and Canada.
 
 
METHODOLOGY
 
The data are the monthly closing indices of the six
KLSE indices, namely, the KLSE industrial index, the
KLSE finance index, the KLSE hotels index, the KLSE
properties index, the KLSE tins index, and the KLSE
plantations index. This study covers a period of 19
years, starting from January 1970 to December 1988.
These indices are widely quoted in Malaysia and
believed to be representative of their respective
sector.
 
The monthly rate of return for month t in year n was
calculated as
 
Rt n = [(It,n-It-t,n)/lt-1,n]
 
where It n and 1~ 1,n refer to monthly closing index
for month t and month t-1, respectively. The average
 
monthly returns for month t were calculated using
the formula
 
n Rt - = E Rt j/n i=1
 
 
where, Rt i refers to the return in month t for year
i, and n is the number of years.
 
The calculations for the monthly returns and the
average monthly returns were made for all six
sectors of the KLSE.
 
1. The following 2 null hypotheses were tested: Ho:
yt = ~2 = = ~12. that is, all average monthly rates
of return are equal.
 
2. Ho: Z , that is, the average rate of return for
January equals the average rate of return for month
i.
 
If the general notion of the weak form of the
efficient market hypothesis is valid for Malaysian
market as suggested by Barnes (1986), Laurence
(1986), Neoh Soon Kean (1985) and Nassir Lanjong
(1983), then the first null hypothesis should be
accepted. That is, in an efficient market the
overall return for all months should not be
significantly different from each other.  The second
hypothesis was aimed at finding out a certain month
which might produce returns significantly different
from January. January might produce the highest
return, but whether or not the returns are
significantly different from any other month is
another question.
 
In testing the first null hypothesis, the oneway
analysis of variance or the F-test (for example, see
Berenson, Levine and Goldstein (1983), and Johnson
and Siskin (1980) was used. The observed value of
the test statistic, F-Observed, can be calculated as
 
Between Groups Mean Square
 
F-Observed Within Groups
 
Within Groups Mean Square
 
The null hypothesis is rejected if F-Observed is
greater than the F-table value at the 5 percent
significance level. Degrees of freedom for F-table
value are C-1 and N-C, where C is the number of
groups (12 in our case), and N is the total number
of observations (228 in our case).
 
In testing the second hypothesis, the t-test for
independent samples (for example. see Johnson and
Siskin (1980), and Mood, Graybill and Boes (1974)
was used. In general, the t-Observed can be
calculated as
 
(x, -xj)/Standard Error
 
where, x1 and x; are the average monthly returns for
month 1 and month i, respectively.
 
There are two ways in which the-standard error can
be computed, depending on whether or not the
variances of the two populations are equal.
Fortunately, we can test the null hypothesis,
variance for group 1 equals variance for group i,
using the F-test. The F-statistic can be calculated
as:
 
F-Observed = S,2/Sj2
 
where, s,2 and Sj2 are the variances of tw v
independent- samples of sizes n1 and n respectively.
 
The value of the F-Observed is compared with the
F-table value with n, - 1 and nj - 1 degrees c
freedom at the 5 percent significance level. If the
variances of the two populations are significant
equal, the standard error can be calculated as
 
{[((nn -1)sn2 + (nj - 1)sj2)/(nr + n;-2)]
 
[1/nr + 1/nj]}1/2
 
where, n1 is the number of monthly returns for month
1, nj is the number of monthly returns fo month i,
s,2 is the variance of monthly returns fo month 1,
and Sj2 is the variance of monthly returns fo month
i.
 
The null hypothesis is accepted if the t-observec is
within plus or minus the t-table value, witt nr + nj
- 2 degrees of freedom at the 5 percen significance
level.
 
If the variances of the two populations are
significantly unequal, the standard error can be
calculated as
 
[sr /nr + sj2/nj] /2
 
 
where all the variables are as described in the
above paragraph: The t-table value is based on the 5
percent significance level, with the number of
degrees of freedom given by the smaller of n, or np
 
Finally, the correlation coefficients are also
computed to find out whether or not the movements
(i.e. monthly returns) in these sectors are
inter-related. The correlation coefficienl between
sector 1 and 2 can be computed as
 
Cov (1,2) r1 2
 
 
[(Var 1 )*(Var 2)]1/2
 
where 'Cov (1,2) is the covariance between monthly
returns of sector 1 and sector 2 and Var 1 and Var 2
refer to variances for sector 1 and sector 2,
respectively.
 
 
FINDINGS
 
Table 1 shows the average monthly returns for all
sectors of the KLSE. As can be seen, the highest
average monthly returns, with the exception of the
hotels sector, are in January. For the hotels
sector,. the highest average monthly t return is in
February, followed by August. The average monthly
returns are positive across all sectors of the KLSE
in February, May, and June. In December the average
monthly returns are comparable to February, May, and
June, except for the tins sector. The average
monthly returns are mostly negative in July and
ovember. Looking more closely at the average monthly
returns for the industrial sector, one can see that
the returns are positive in all months from December
to July. In the plantations sector the positive
returns are in the months of December through June
whereas in the tins sector, the positive returns are
in January through June. The average monthly returns
are mixed in other sectors.
 
Table 2 shows the results of the F-test according to
sector. As indicated by the F-statistic values and
the P-values, none of the sectors exhibits
significant overall difference in terms of their
average monthly returns at the 5 percent level. This
finding supports the weak form of the efficient
market hypothesis (see Fama (1970)) which contends
that all months, more or less, provide equal
returns. Actually, the Scheffe techniques of
multiple comparison (see Berenson and Goldstein
(1983)) was also applied to further investigate the
results of the F-test. The Scheffe technique did
confirm the finding of the F-test at the 5 percent
level of significance.
 
The results of the t-test are shown in Table 3. With
the exception of the months of February, May, June
and December, all other months seem to have average
returns significantly different from January at the
5 percent level for the industrial sector. However,
only August, September, and November do differ
significantly from January in terms of the average
monthly return at the 1 percent level. For the
finance sector, only the average return in March
differs significantly from January at the 5 percent
level, and none differs significantly at the 1
percent level. In the hotels sector, no month is
significantly different from January in terms of the
average monthly return at the 5 percent level. In
the properties sector, with the exception of
September and November, the average returns of all
months are not significantly different from January,
and none differs at the 1 percent significance level
in the tins sector, the difference occurs for
September and November at both the 5 and 1 percents
significance level. In the plantations sector, the
differences are for the months of July, August,
September, and November at the 5 percent
significance level, and for the months of July,
August, and September at the 1 percent significance
level. When all other months are combined, ony
hotels sector exhibits no significant difference at
the 5 percent level. Only the industrial sector
shows significant difference at the 1 percent level.
 
Contradictory results between the F-test and the
t-test should be explained here. Norusis (1983, p.
111) pointed out that a significant F-statistic
indicates only that the population means are
probably unequal without pinpointing where the
differences are. A variety of special techniques,
termed multiple comparison procedures, such as
Scheffe test, can be used to determine which
population means are different from each other.
These multiple comparison procedures protect against
calling too many differences significant. These
procedures set up more stringent criteria for
declaring differences significant than does the
usual t-test. That is, the difference between two
sample means must be larger to be identified as a
true difference. Norusis also mentioned that
Snedecor and Cochran (1967) stated that there is a
problem when t-test is used to test all possible
pairs of means. The problem is that when many
comparisons are made, some will appear to be
significant. even when all population means are
equal.  With five groups, for example, there are ten
possible comparisons between pairs of means. When
all population means are equal, the probability that
at Ieast one of the ten observed significance levels
will be less than 0.05 is about 0.29.
 
The correlation coefficients of the monthly returns
between sectors are shown in Table 4. All
correlation coefficients are significant at the 5
percent level. The correlations between the
industrial sector and the.other sectors, with the
exception of the hotels sector (correlation equals
0.4328), are high. The correlations between hotels
sector and tins and plantations sectors are quite
low, i.e., 0.2511 and 0.3672 respectively. The
correlations between the rest of the sectors are
somewhere between 0.4817 and 0.7749.
 
 
CONCLUSIONS AND IMPLICATIONS
 
Overall, initial observation shows that there exists
a month, !.e. January, which consistently produced
the highest return in 5 of the 6 sectors of the
KLSE. The hotels sector produced the highest return
in February and the second highest return in August.
In general,- positive returns can be expected in the
months of December through February. This initial
observation seems to conform to the January' and
end-of-the-year effects. The results of the t-test
somewhat reinforce the existence of these effects in
the Malaysian market, most notably in the industrial
sector. In the hotels sector, the return in January
is not significantly different from any other month,
which is something to be expected due to mostly
inactive trading in this sector compared to other
sectors. It should be noted here that the trading in
the hotels sector is mostly on the Singaporean
stocks. In other sectors, only a few months seem to
differ from January in terms of the average monthly
return.
 
If one were to accept the existence of the January
and end-of-the year effects in Malaysia, then one
has to offer some possible explanations for this
phenomenon. In the West, one can say that it might
be due to taxes on the capital gains, but in
Malaysia there is no tax whatsoever on capital
gains. Another explanation is; professional
investors (portfolio managers) move out (i.e. sell)
the re from the more risky stocks portfolio to less
risky stocks portfolio at the end of the year to
give a more conservative picture of the portfolios
which they manage. This action drives the stock
prices down. In January, they move back into the
market, thus drives the prices up. This might be
true in Malaysia.
 
In Malaysia, most of the participants in the stock
market are the Chinese. They celebrate Chinese New
Year in February on a grand scale, with the giving
of "angpow" (gifts, normally cash money) to friends
and relatives. One way of getting money is through
speculation in the stock market. Therefore, the
speculators start moving into the market as early as
December, thus driving the prices up. On the other
hand, the prices start to lose momentum in February
when most of these speculators move out of the
market. In Malaysia, this phenomenon is called the
Chinese New Year effect.
 
As discussed earlier in the previous section, the
F-test for multiple comparison is more strict than
the t-test in its decision to reject the null
hypothesis of no difference in means. For one thing,
the efficient market hypothesis is a general
concept, and therefore, a general test such as the
F-test is somewhat more appropriate to test the
validity of the hypothesis.  Furthermore, the
author's - calculations of the skewness of these
returns show that the distribution of the monthly
returns are skewed to the right, which is consistent
with the F-Distribution. This means that the market
is still efficient in the weak form even with the
seemingly high return in January. In other words,
the efficient market hypothesis (the weak form) is
still valid.
 
The results of the correlation coefficients
indicated that the inter-relatedness among sectors
does exist.  This means that the events in one
sector do affect the other sectors. It also mean
that the same events in the country (i.e., macro
events) do affect all the sectors in the KLSE.
 
 
SELECTED REFERENCES
 
Anuar Nasir, and Shamsher Mohamad. 198 (August),
"Stock Returns and the Weekend Effect The Malaysian
Experience." Proceeding. Academy of international
Business, South East Asia Regional Conference, Putra
World Trade Center, Kuala Lumpur Malaysia, V70 -
V84.
 
Barnes, Paul. 1986 (Winter). "Thin Trading and Stock
Market Efficiency: The Case of The Kuala Lumpur
Stock Exchange." Journal of Business and Accounting,
609 - 617.
 
Berenson, M., Levine, D., and Goldstein, M. 1983
Intermediate Statistical Methods and Applications A
Computer Package Approach. Prentice-Hall Englewood
Cliffs.
 
Berges, A., McConnell, J. J., and Schlarbaum, G G.
1984 (March). "An Investigation of the turn-of
the-year Effect, the Small Firm Effect and the Tax
Loss Selling Pressure Hypothesis in Canadian Stock
Returns. " Journal of Finance 39,185 - 192.
 
Brown, P., Keim, D. B., Keildon, A. W., and Marsh,
T. A.  1983. "Stock Return Seasonalities and the
Tax-Loss Selling Hypothesis - Australian Evidence."
Journal of Financial Economics 12, 105 - 127.
 
D'Ambrosio, C. A. 1980 (Spring). " Random Walk and
The Stock Exchange of Singapore." Financial Review
15, 1 - 12.
 
Dyl, Edward A. 1977 (March). "Capital Gain Taxation
and Year End Stock Market Behavior." Journal of
Finance 32.
 
Fama, E.F. 1965 (January) "The Behavior of Stock
Market Prices." Journal of Business, 34 - 105.
 
Fama, E.F. 1970 (March). "Efficient Capital Markets:
A Review of Theory and Empirical Works." The Journal
of Finance, 383 - 417.
 
Gibbons, M. R., and Hess, P.1981 (October). "Day of
the Week Effects and Asset Returns." Journal of
Business 54, 579 - 596.
 
Givoly, Dan, and Ovadia, Arie. 1983. "Year-End
Tax-lnduced Sales and Stock Market Seasonality."
Journal of Finance 38, 171 - 185.
 
Gultekin, M. N., and Gultekin, B. N. 1983
(December).  "Stock Market Seasonality:
International Evidence." Journal of Financial
Economics 12, 469 - 481.  I Hong, H. 1978
(November). "Predictability of Price Trends on Stock
Exchanges: A Study of Some Far Eastern Countries."
Review of Economics and statistics, 619 - 621.
 
Johnson, R. and Siskin, 8. (1980). Elementary
Statistics for Business. Duxbury Press, North
Scituate, Mass.
 
Keim, Donald B. 1983. ~Size-Related Anomalies and
Stock Return Seasonality". Journal of Financial
Economics 12.
 
Laurence, M. 1986. "Weak Form Efficiency in the
Kuala Lumpur and Singapore Stock Exchanges." Journal
of Business and Finance 10, 431 - 445.
 
Mood, A., Graybill, F., and Boes, D. 1974.
Introduction to the Theory of Statistics, 3rd ed.
McGraw-Hill, New York.
 
 
Nassir Lanjong, 1983. "Efficiency of the Malaysian
Stock Market and Risk-Return Relationship."
Unpublished Ph.D. Thesis, Leuven.
 
Neoh Soon Kean, 1985. "An Examination of the
Efficiency of the Malaysian Stockmarket."
Unpublished Ph.D. Thesis, Edinburgh.
 
Neter, J., Wasserman, W., and Kutner, M. 1985.
Applied Linear Statistical Models, 2nd ed.*Richard
D. Irwin, Homewood.
 
Norusis, M. J. 1983. Introductory Statistics Guide:
SPSS-X.  McGraw-Hill, New York.
 
Roll, R.1982 (March). "The-Turn-of-Year Effect and
the Return Premia of Small Firms." Graduate School
of Management, UCLA. Revised.
 
Rozeff, Michael S., and Kinney, William R. Jr.1976.
"Capital Market Seasonaiity: The Case of Stock
Returns". Journal of Financial Economics 3, 379 402.
 
Praetz, P. D. 1973. " Analysis of Australian Stock
Prices." Australian Economic Papers 20, No. 12.
 
Snedecor, G. W. and Cochran. 1967. Statistical
Method. Iowa State University Press, Ames.
 
Tinic, S. M., and West, R. 1984 (December), "Risk
and Return: January versus the Rest of the Year."
Journal of Financial Economics 13, 561 - 574.