CHAPTER 5 FINDINGS OF THE STUDY -...
Transcript of CHAPTER 5 FINDINGS OF THE STUDY -...
352
CHAPTER 5
FINDINGS OF THE STUDY
The research mainly aimed at understanding;
i. Whether the selected indices confirm the existence of a certain anomaly?
ii. If anomalies exist, whether these anomalies are stable and consistent over time
and across indices considered i.e., are they true anomalies?
iii. Whether these anomalies observed, show any consistency with the existing
literature.
The major findings for each calendar anomaly observed from the econometric analysis
are summarized below:
5.1 Day-of-the-week effect results
The null hypothesis tested the absence of day-of-the-week seasonality in the BSE and
NSE indices. The day-of-the-week effect states that the daily average index returns in
the framework of an efficient market should be uniformly distributed over different days
of the week. The rejection of the null hypothesis points towards presence of a pattern in
the data and thus coefficients of the dummy variables are significantly different from
zero. The most often referred anomaly is the weekend effect, wherein the mean returns
observed on Monday should be significantly lower (negative) while the mean returns on
Friday has to be significantly higher (positive) in nature. The results of the econometric
analysis for all the indices are summarized in Table-181 to 183. These charts summarize
the significant weekdays considering standard OLS model, GARCH model, and
GARCH-M model respectively for sub-periods (pre- and post- 2002)1 and overall period
at one percent and five percent significance levels. The Table-181 provides the summary
of only key indices listed on BSE and NSE respectively namely BSE Sensex, CNX
Nifty and CNX Nifty Junior index. At first, considering the overall period, we notice
significant similarities in all three indices. The mean return is observed to be the highest
1 For the summary of the results, we considered two sub-periods for only those indices with data availability of minimum three years
before 2002.
353
on Wednesday consistently for NSE listed indices in all three models, whereas in case of
BSE Sensex, there seems to be no consistency (Woolridge (1991) and Lumsdaine and
Ng (1999))2 wherein Wednesday was observed to have the highest mean returns in
regression model, but, Friday and Monday were observed to have highest returns after
considering time varying volatility.3 The lowest return for all the three indices was
observed on Tuesday respectively4. Thus, Wednesday effect and Monday effect can be
considered to be weekdays of truly anomalous returns in the NSE indices and BSE
Sensex index respectively and thus Wednesday effect and Monday effect cannot be
explained by time varying volatility. Following Wang et.al (1997), the results of the sub-
period analysis points towards change in the stability of the day-of-the-week effect due
to structural changes which have taken place in the Indian capital markets during the
period, 1990 to 2011. In the sub-period I, it is observed that Wednesday being the
weekday with highest mean returns followed by Tuesday having the lowest returns in all
the indices considering the time varying market volatility. Whereas in sub-period II,
significantly higher returns on Friday and the lower returns on Tuesday are noticed, after
considering the time varying volatility for BSE Sensex and CNX Nifty Junior index5.
The Tables 182 and 183 present the results of day-of-the-week effect in returns and
volatility of BSE and NSE listed broader and sectoral indices. At first, considering the
broader indices mainly the mid-cap and small-cap indices for the overall period, we can
observe that Monday remains to be significant in all the three models except for NIFTY
Mid-cap50 index. Thus Monday effect can be considered as true anomaly. Monday and
Tuesday are observed to have the highest and the lowest returns for the overall period.
Only in case of NIFTY Mid-cap50 index, we can observe that Friday and Tuesday have
the highest and lowest returns respectively. Considering all the sectoral indices, Monday
appears to remain significant in all three models and hence can be depicted to be true
anomaly. In case of all sectoral indices except for S&P BSE TECK and NSE sectoral
indices, Monday mean returns are observed to be significantly higher when compared to
2 Infer that the previous results obtained on Indian capital markets are mainly based on failure of research papers to model the
conditional mean which inturn lead to erroneous conclusions. 3 The results deviate from the studies by Fields (1930), French(1980), Fama (1965) etc,. where they observed positive Friday returns
and negative Monday returns (weekend effect). 4 Jaffe and Westerfield (1985), Peiro (1994), Agarwal and Tandon (1994) point towards different patterns i.e., either Monday or
Tuesday having the least returns. 5 Lakonishok and Smidt (1988) and Agarwal & Tandon (1994), point towards disappearance of weekend effect gradually in many
countries.
354
other weekdays and Tuesday remains to have the lowest mean returns. Thus, the result
of the econometric models shows high positive returns on Wednesday and Monday for
broader indices and sectoral indices respectively. The largest mean returns on Monday
especially for sectoral indices (when compared to higher returns of Friday and lowest
returns on Monday as observed in developed countries) point towards lagging effect of
Indian sectors taking cues from the global markets. The results observed point towards
“wait and watch principle” followed by investors. These high returns towards the
beginning of the week followed by lowest returns on Tuesday is in contrast to the
evidence obtained from other markets. In case of broader indices especially BSE
Sensex, CNX Nifty and CNX Nifty junior index, higher significant returns are obtained
on Wednesday but with lowest returns on Tuesdays.6 Thus, a pattern of higher Mondays
and higher Wednesdays can be observed with intermittent lower Tuesdays. Thus,
investors can adopt buy- hold- sell strategy in case of broader indices and sell-hold-buy
strategy in case of sectoral indices to make abnormal gains. We also observed
previously that in case of Indian markets, the returns obtained seem to be particularly
higher in the first sub-period examined when compared to returns for the total time
period. The last sub-period examined in all the indices present lower returns which are
infact closer to those observed in developed countries.7 Moreover, the results indicate
that the Tuesday returns are significantly negative and have become still smaller in the
recent periods8. This trend of negative returns on Tuesday has also been noticed in other
markets too.9 Thus in trying to explain these findings for Indian stock market, it could
be argued that investors take a optimistic road in the beginning of the week taking cues
from international markets on Monday10
. Karmakar and Chakraborty (2003), Nath and
Dalvi (2004) , Selvarani and Jenefa (2009), Garg and Chhabra (2010) find presence of
day anomalies in the Indian stock market which in turn are high correlated with the
6 Deepa Mangala (2008), studied day-of-the-week effect on CNX Nifty index during the period 1991 to 2007. The study also found
highest positive returns on Wednesday and most negative returns on Tuesday for the entire period. 7 Bafna and Mathur (2010) find no significant presence of the day-of-the-week effect in BSE Sensex returns for the period 2000-
2010. 8 Deepa Mangala (2008), attribute the presence of high returns on Wednesday and lowest returns on Tuesday mainly to
introduction of rolling settlement (from Tuesday settlement) in NSE. 9 Jaffe and Westerfield (1985a), Condoyanni et al. (1987) observed negative returns on Tuesday in Singapore and France, While in
Australia and Japan, both Tuesday and Mondays were observed to have negative returns. 10
Lakonishok and Maberly (1990) believed weekend effect can be explained by participation of more retail investors in the market
during Monday when compared to institutional investors. Thus more trading by retail investors (especially selling) on Mondays may be the cause for weekend effect.
355
investment pattern of FIIs which also invest in the similar way. The studies also find the
day anomalies persistent in the recent years but slowly losing its consistency, thereby
reconfirming that various trading patterns adopted would not be profitable with time.
The studies also find net investments by FIIs to follow the pattern of Friday effect. After
confronting with the realities of Indian economy and understanding the performance of
sectoral indices, investors are observed to react in the broader index such as BSE
Sensex, CNX Nifty and CNX Nifty Junior which are nonetheless the combinations of
major sectors. Thus, the results indicate that investors can adopt varying trading
strategies dealing with sectoral indices and broader indices separately. The results
indicate that the day-of-the-week effect is persistent in Indian stock market and is
unique to Indian market. Thus, the findings reject the null hypothesis for presence of
day-of-the-week effect in Indian stock markets. In both broader and sectoral indices, the
anomalies appear to be stable in nature. But, the day-of-the-week effect seems to be
disappear with time as the margin of returns is observed to decline11
.
11
Bhattacharya, K., Sarkar, N. and Mukhopadhayay, D. (2008) obtained significant changes in returns and volatility since mid-
1990’s. They suggested possible interaction between banking sector and capital market could explain day-of-the-week effect in returns and inter-exchange arbitrage oppurtunities due to difference in settlement period could explain seasonality in volatility.
356
Table 181: Day of the week effect in returns and market volatility of BSE and NSE listed indices for the overall period and sub-periods
Source: Author Note:
The terms “Mon, Tue, Wed, Thur, Fri” refer to the days of the week from Monday to Friday respectively
*Significance at 5 percent level
** Significance at 1 percent level
357
Table 182: Day of the week effect in returns and market volatility of BSE and NSE listed indices for the overall period and sub-periods
(contd..)
Source: Author
Note: The terms “Mon, Tue, Wed, Thur, Fri” refer to the days of the week from Monday to Friday respectively
*Significance at 5 percent level
** Significance at 1 percent level
358
Table 183: Day of the week effect in returns and market volatility of BSE and NSE listed indices for the overall period and sub-periods
(contd..)
Source: Author
Note: The terms “Mon, Tue, Wed, Thur, Fri” refer to the days of the week from Monday to Friday respectively
*Significance at 5 percent level
** Significance at 1 percent level
359
5.2 Month-of-the-year effect results
According to Efficient Market hypothesis, returns should be equal among months and
there should be existence of no seasonal patterns between months. Thus the null
hypothesis of mean daily returns being statistically not different among months was
tested. Any violation would inturn reflect on possibility of market irregularity. In
literature, the existence of January effect is documented many times, which states that
January returns are consistently higher than the returns of any other month. There is
found to be validation for the presence of the month-of-the-year effect in some of the
examined indices during the entire sample period. The Tables 184 to 187 present the
results with respect to month-of-the-year effect in returns and volatility of all the indices
obtained from standard OLS model, GARCH model and GARCH-M models
respectively. The study reveals significant results regarding month-of-the-year effect in
Indian stock market at both broader market level and sectoral level respectively. At first,
we consider the results obtained in case of broader indices mainly BSE Sensex, CNX
Nifty and CNX Nifty Junior index as shown in Table-184. For the overall period,
significant December effect is observed in the return equation of both CNX Nifty and
CNX Nifty Junior indices even after considering the time varying market volatility in
the index returns. December is observed to have the highest returns followed by October
having the least returns. Whereas, January effect is observed in case of BSE Sensex
index considering the overall period.12
January remains to be significant in all the three
models for BSE Sensex implying that January effect is a true anomaly. However,
February effect is found no more significant under GARCH estimations but remains
significant in the returns volatility. Considering the OLS estimations, December and
October month were observed to have highest and lowest average returns respectively in
both CNX Nifty and CNX Nifty Junior index whereas, February and April months had
highest and lowest average returns in case of BSE Sensex. Considering time varying
volatility, we observe returns to be highest in May month for BSE Sensex whereas
December month retains highest returns in case of NSE indices.
12
Kaur (2004) also found higher positive returns in February and December month but did not find evidence of January effect in
Indian stock markets. Bodla and Jindal (2006) found no significant differences in the monthly Sensex returns. Fountas and segredakis (2002) found August returns to be significantly higher than april, may, October and November returns.
360
Considering the sub-periods, first sub-period examined shows presence of
significant February effect which seem to have changed pattern in second sub-period
wherein we observe significant December effect in both NSE indices but June effect in
case of BSE Sensex index13
. Pandey (2002) observed March effect in BSE Sensex for
the period 1991-2001, attributing the results to tax-loss hypothesis (Wachtel, 1942)
which are similar to our results in the first sub-period. But the patterns are observed to
change in the second sub-period and hence tax-loss hypothesis cannot be considered as
an explanation to month-of-the-year effect. Considering the results of mid-cap and
small-cap indices presented in Table-185, we can infer that April effect and December
effect remain significant in the return equation of all the three models for the overall
period and hence can be considered to be true anomalies. Combining the results of all
the three models, December is observed to have the highest returns followed by lowest
returns in January in case of BSE Mid-cap and BSE Small-cap indices, whereas in case
of NIFTY Mid-cap50 index, the highest returns were observed in April followed by
lowest returns in January14
.
Finally, considering the results of sectoral indices as presented in Tables 185 to
187, we can observe that, the coefficients for the months of August, November and
December are significant in the mean equation even after considering the time varying
volatility except for in case of S&P BSE Teck, S&P BSE O&G and CNX Infra index
respectively. Whereas, none of the months were observed to be significant in the
Standard OLS model. After considering the time varying volatility, the November and
December month mean returns are observed to be significantly higher compared to all
the months followed by January having the lowest returns in majority of the indices.
Thus, from the analysis we can notice that, the mean monthly returns are significantly
different from zero mainly in the Months of January, February, and December. The
higher Positive December mean returns followed by negative returns in the months of
January and February could be caused by a change in investor’s behavior, anticipating
January effect and March effect in Indian stock markets and other stock markets since
13
Yakob, Beal and Delpachitra (2005) considered period 2000 to 2005 as a period free of influence from asian crisis and thus a
stable period for studying seasonality. They found highest positive returns in November, and significantly lowest negative returns
in April. 14
Roll (1983) found significant differences in the returns patterns of large and small firms, especially around turn-of-the-year. Small
firm premium increases between December month of previous year and first three months of the next year.
361
the fiscal year in India starts in April and ends in March, whereas it is January to
December in other developed countries. Furthermore, a closer look at the sub-period
values reveals that, December month is statistically significant in recent period i.e.,
2002-2011. The significant February effect observed in Indian markets in the first sub-
periods seems to have changed. Thus from the analysis, we can conclude that though tax
loss hypothesis helps explain monthly effect in Indian stock markets for a brief period,
but all the indices indicate a disappearing March/April effect over the whole sample
period. Even after considering the time-varying volatility, the results reconfirm the OLS
regression results. December is found to be very significant in all the broader indices
except for BSE Sensex index wherein January is observed to be significant. whereas in
case of sectoral indices, December was observed to be significant for all the indices
except for BSE FMCG, BSE Teck, BSE O&G and CNX INFRA Indices respectively.
Thus the results obtained indicated higher integration of Indian markets than ever in the
recent period15
.
15
Schwert (2002) , Sharma & Kennedy (1977) pointed towards the disappearance of anomalies over time maybe due to rational trader exploiting the documented anomalous behavior, hence leading to efficient markets.
362
Table 184: Month of the year effect in returns and market volatility of BSE and NSE listed indices for the overall period and sub-periods
Source: Author
Note: The terms “Jan, Feb, Mar, Apr, Aug, Sept, Nov, and Dec” refer to the months of the year from January to December respectively
*Significance at 5 percent level
** Significance at 1 percent level
363
Table-185: Month of the year effect in returns and market volatility of BSE and NSE listed indices for the overall period and sub-periods
(contd..)
Source: Author Note: The terms “Jan, Feb, Mar, Apr, Aug, Sept, Nov, and Dec” refer to the months of the year from January to December respectively
*Significance at 5 percent level
** Significance at 1 percent level
364
Table-186: Month of the year effect in returns and market volatility of BSE and NSE listed indices for the overall period and sub-periods
(contd..)
Source: Author Note: The terms “Jan, Feb, Mar, Apr, Aug, Sept, Nov, and Dec” refer to the months of the year from January to December respectively
*Significance at 5 percent level
** Significance at 1 percent level
365
Table-187: Month of the year effect in returns and market volatility of BSE and NSE listed indices for the overall period and sub-periods
(contd..)
Source: Author
Note: The terms “Jan, Feb, Mar, Apr, Aug, Sept, Nov, and Dec” refer to the months of the year from January to December respectively
*Significance at 5 percent level
** Significance at 1 percent level
366
5.3 Holiday Effect
The Tables 188 to 191 show the mean (daily) index return for days around important
holidays and normal days for the overall period and sub-periods. First, the daily rate of
return around important holidays for major indices namely BSE Sensex index, CNX
Nifty index and CNX Nifty Junior index is presented in Table-188. The results point
towards disappearance of the holiday effect in the recent periods in these indices. For
the overall period, considering BSE Sensex index, the mean returns around Ganesh
Chaturthi and Bakrid holidays were found to be significantly different from other
normal trading days. Whereas, the mean returns around Dussehra and Mohurram
holidays were significant for CNX Nifty and CNX Nifty Junior indices. All the
coefficients for the above said holidays were observed to be negative in nature. On
average, the returns around Dussehra and Mohurram holidays were observed to around
five times lower when compared to mean returns around non-holiday periods. The
results obtained for the overall period seems to match the results obtained in the first
sub-period. But, disappearance of holiday effect in the second sub-period for these
indices is observed which points towards over-exploitation of the holiday effect by the
investors due to more awareness on these anomalies. It should also be noticed that the
average returns are lower when the analysis refers to the total time period, while the last
sub-period presents even lower returns which come closer to those observed in other
developed markets.
The Tables 189 to 191 presents the summary of the daily rates of return around
important holiday for sectoral indices. Considering mid-cap and small-cap indices for
the overall period, mean returns around Dussehra, Diwali and Mohurram are found to be
significantly different from non-holiday period. Negative coefficients are observed
around Dussehra and Mohurram holidays, whereas positive mean returns can be
observed around Diwali holidays. On an average, the mean returns around Dussehra and
Mohurram holidays were observed to be seven times lower when compared to mean
returns around non-holiday period. French(1980) suggested holiday returns to be
negative in nature based on the argument that the depressed Monday returns from
weekend effect are due to market closing. Whereas, the mean returns around Diwali
367
holidays was observed to be around ten times higher (Ariel,1990)16
. Further, in case of
sectoral indices also, same results re-emerge with the mean returns around Dussehra,
Diwali, Bakrid and Mohurram holidays to be significant in nature. The mean returns are
much lower (around six times lower when compared to non-holiday period) around
Dussehra, Bakrid and Mohurram holidays. Whereas, the mean returns around Diwali
holidays is significantly higher (around ten times more than mean returns around non-
holiday period). Though in the first sub-period same results were obtained, we can
notice that the holiday effect seem to disappear in second sub-period examined for all
the indices. Holiday effect seems not to be present in few indices mainly BSE CD, BSE
IT, BSE PSU, BSE AUTO, CNX IT respectively. In rest other indices, The mean returns
around major holiday mainly Ganesh chaturthi, Dussehra, Diwali, Bakrid and
Mohurram holidays are observed to be significant in nature. Thus we can observe
holiday effect to be present in all firm sizes and are more pronounced in the broader
indices than sectoral indices (Kim and Park (1994) found the same results for US
markets). Holidays were further examined based on their occurrence on weekdays and
towards specific half of the calendar month. For the study, we considered 207 holiday
dates spread across 21 years from 1990 to 2011 for nine holidays. At first, considering
the occurrence of holidays on weekdays, we find very interesting results. The results are
summarized in Figure 5 and Table-A7. In the order of the occurrence of Holidays on
weekdays, Wednesday and Friday are observed to be the days highly likely to have
holidays than rest of the other days. If we sub-divide the calendar year based on
occurrence of holidays, important holidays such as Ganesh chaturthi, Dussehra, Diwali
and Bakrid occur mainly during the later part of the calendar year i.e. between
September and December. Especially, these holidays seem to occur mainly in the second
half of the month as shown in Figure 6 and Table-A8.
Similarly, other holidays namely Mohurram, Maha Shivaratri and Rama Navami
usually occur during January and April months. Considering these holidays, the
percentage of occurrence of holidays on Tuesday is found to be highest (27%) followed
by Wednesday (20%). The holidays namely Maha Shivaratri, Rama Navami, Ramzan
16
found mean returns around pre-holidays to be significantly higher and more pronounced in nature.
368
are observed to occur towards the first half of the calendar month during the entire study
period. Thus, there is very likely chance that holiday effect is the reason for semi-month
effect and Wednesday effect as the behavior of Wednesday return behavior is found to
be dissimilar to security returns around holidays (Ariel, 1990).
Figure 5: Occurrence of major holidays on trading days of exchanges (%)
Source: Author
Figure 6: Occurrence of holidays during first half and second half of the months
for the overall study period (1990-2011)
Source: Author
8
9
10
11
12
13
14
FIRST HALF OF MONTH SECOND HALF OF THE MONTH
MAHA SHIVARATRI
RAMA NAVAMI
RAMZAN ID
GANESH CHATURTHI
DUSSERA MAHANAVAMI
DIWALI
BAKRID
GURU NANAK JAYANTHI
MOHURRAM
369
Table 188: Daily rate of return of BSE and NSE listed indices around important holidays for the overall period
Source: Author
Note:
* *and * denote Significance at 1 and 5 percent level. Coefficients of Holidays are reported in each cell. Just below the coefficients, the numbers depict t-statistics.
OTHER DAYS
0.092 -0.180 0.124 -0.313 * -0.231 -0.249 -0.349 * -0.035 0.208 0.054
0.587 -1.141 0.781 -1.985 -1.470 -1.121 -2.214 -0.205 1.319
0.095 -0.277 -0.004 -0.486 * -0.492 * -0.545 -0.534 ** 0.104 0.091 0.041
0.423 -1.231 -0.017 -2.081 -2.116 -1.688 -2.388 0.375 0.428
0.118 -0.014 0.210 -0.162 0.024 0.038 -0.114 -0.170 0.368 0.070
0.537 -0.061 0.968 -0.766 0.112 0.125 -0.512 -0.805 1.563
0.107 0.184 0.234 0.203 -0.362 * 0.301 0.103 -0.099 -0.419 ** 0.049
0.663 1.138 1.391 1.257 -2.244 1.300 0.617 -0.597 -2.522
0.403 0.435 0.048 0.042 -0.352 -0.012 -0.029 -0.261 -0.552 * 0.008
1.595 1.693 0.191 0.178 -1.512 -0.036 -0.113 -1.050 -2.200
-0.080 0.038 0.316 0.332 -0.406 0.607 0.150 0.029 -0.344 0.080
-0.380 0.179 1.378 1.469 -1.807 1.873 0.669 0.130 -1.538
-0.115 0.284 0.026 0.202 -0.325 0.512 0.000 -0.068 -0.563 ** 0.072
-0.598 1.469 0.128 1.015 -1.643 1.847 0.000 -0.352 -2.835
0.025 0.590 -0.297 0.287 -0.313 0.205 -0.320 -0.237 -0.824 ** 0.023
0.077 1.713 -0.890 0.863 -0.963 0.468 -0.938 -0.775 -2.473
-0.181 0.168 0.246 0.113 -0.364 0.745 * 0.143 0.060 -0.431 0.104
-0.772 0.719 0.967 0.452 -1.463 2.074 0.576 0.242 -1.740
MAHA
SHIVARATRI
RAMA
NAVAMI
OVERALL PERIOD
(1995-2011)
OVERALL PERIOD
(1997-2011)
BSE SENSEX
CNX NIFTY
CNX NIFTY JUNIOR
Significant HolidaysIndicesoverall period/sub-
periods
RAMZAN IDGANESH
CHATURTHI
DUSSERA
MAHANAVAMIINDICES overall period/sub-periods
OVERALL PERIOD
(1991-2011)
DIWALI BAKRID GURU NANAK
JAYANTHIMOHURRAM
SUB-PERIOD I
(1991-2001)
SUB-PERIOD II
(2002-2011)
SUB-PERIOD I
(1995-2001)
SUB-PERIOD II
(2002-2011)
SUB-PERIOD I
(1997-2001)
SUB-PERIOD II
(2002-2011)
370
Table 189: Daily rate of return of BSE and NSE listed indices around important holidays for the overall period (contd..)
Source: Author
Note: * *and * denote Significance at 1 and 5 percent level. Coefficients of Holidays are reported in each cell. Just below the coefficients, the numbers depict t-
statistics.
OTHER DAYS
-0.276 0.197 0.202 0.109 -0.484 * 0.876 ** 0.076 0.083 -0.540 * 0.109
-1.172 0.887 0.827 0.455 -2.034 2.526 0.299 0.350 -2.133
-0.371 0.247 0.210 0.056 -0.521 * 1.079 ** 0.015 0.081 -0.560 * 0.122
-1.493 1.055 0.815 0.220 -2.075 2.947 0.057 0.321 -2.095
-0.264 0.237 0.157 0.142 -0.663 * 1.021 ** 0.207 -0.063 -0.764 ** 0.066
-0.943 0.845 0.499 0.461 -2.182 2.359 0.686 -0.208 -2.535
-0.138 0.411 0.140 -0.004 -0.413 0.602 0.045 -0.033 -0.378 0.099
-0.625 1.859 0.594 -0.018 -1.790 1.802 0.197 -0.144 -1.639
0.125 0.636 -0.246 -0.231 -0.086 0.635 -0.772 0.065 -1.199 ** -0.051
0.272 1.311 -0.522 -0.480 -0.189 0.939 -1.657 0.141 -2.508
-0.215 0.450 0.254 0.049 -0.527 * 0.622 0.286 -0.067 -0.147 0.145
-0.854 1.789 0.930 0.183 -1.971 1.613 1.072 -0.249 -0.552
-0.079 0.268 0.223 0.089 -0.317 0.220 -0.597 ** 0.103 -0.505 * 0.082
-0.331 1.121 0.879 0.353 -1.272 0.607 -2.395 0.414 -2.024
0.158 0.573 0.346 0.295 0.320 0.554 -1.419 ** 0.067 -1.195 * -0.037
0.290 0.994 0.616 0.516 0.587 0.691 -2.567 0.122 -2.105
-0.157 0.273 0.202 -0.010 -0.523 0.107 -0.357 0.125 -0.325 0.118
-0.597 1.035 0.705 -0.036 -1.867 0.266 -1.280 0.447 -1.165
-0.045 -0.075 0.093 0.063 -0.215 0.507 * -0.090 0.147 -0.310 0.053
-0.264 -0.442 0.517 0.356 -1.214 1.978 -0.508 0.830 -1.756
0.536 0.360 0.569 -0.045 -0.658 -0.233 -0.235 0.157 -0.984 * 0.005
1.284 0.817 1.327 -0.103 -1.582 -0.380 -0.555 0.377 -2.267
-0.233 -0.149 -0.091 0.145 -0.066 0.743 ** -0.081 0.126 -0.127 0.068
-1.291 -0.824 -0.464 0.751 -0.345 2.682 -0.426 0.659 -0.665
DIWALI BAKRID GURU NANAK
JAYANTHIMOHURRAM
SUB-PERIOD II
(2002-2011)
S&P BSE MID-CAP OVERALL PERIOD
(2005-2011)
SUB-PERIOD II
(2002-2011)
S&P BSE CG SUB-PERIOD I
(1999-2001)
Indicesoverall period/sub-
periodsSignificant Holidays
INDICES overall period/sub-periodsMAHA
SHIVARATRI
RAMA
NAVAMIRAMZAN ID
GANESH
CHATURTHI
DUSSERA
MAHANAVAMI
OVERALL PERIOD
(2005-2011)
OVERALL PERIOD
(2005-2011)
OVERALL PERIOD
(1999-2011)
S&P BSE SMALL-CAP
CNX MID-CAP50
OVERALL PERIOD
(1999-2011)
SUB-PERIOD I
(1999-2001)
SUB-PERIOD II
(2002-2011)
S&P BSE CD
S&P BSE FMCG
OVERALL PERIOD
(1999-2011)
SUB-PERIOD I
(1999-2001)
371
Table 190: Daily rate of return of BSE and NSE listed indices around important holidays for the overall period (contd..)
Source: Author
Note: * *and * denote Significance at 1 and 5 percent level. Coefficients of Holidays are reported in each cell. Just below the coefficients, the numbers depict t-
statistics.
OTHER DAYS
0.212 0.313 -0.067 0.176 -0.371 * 0.325 -0.045 0.223 -0.705 ** 0.068
1.280 1.879 -0.378 1.009 -2.138 1.294 -0.257 1.284 -4.066
0.915 * 0.838 -0.848 0.165 -0.552 0.007 -0.499 0.519 -1.860 ** 0.049
1.926 1.672 -1.740 0.332 -1.166 0.011 -1.038 1.097 -3.769
0.007 0.266 0.185 0.094 -0.314 0.445 0.054 0.112 -0.378 * 0.074
0.041 1.653 1.061 0.547 -1.835 1.802 0.315 0.655 -2.220
-0.192 -0.263 0.112 0.443 -0.611 * 0.399 -0.592 * 0.566 -0.699 * 0.091
-0.646 -0.881 0.352 1.417 -1.961 0.884 -1.905 1.821 -2.246
-0.851 1.079 -0.142 0.346 -1.551 0.363 -1.992 * 1.485 -3.043 ** 0.143
-0.908 1.090 -0.147 0.353 -1.658 0.264 -2.098 1.588 -3.123
-0.011 -0.392 0.191 0.341 -0.257 0.394 -0.238 0.201 -0.074 0.075
-0.041 -1.492 0.670 1.215 -0.920 0.978 -0.855 0.721 -0.267
-0.107 0.140 0.256 0.161 -0.221 0.506 -0.015 -0.022 -0.386 0.083
-0.511 0.670 1.152 0.734 -1.015 1.602 -0.069 -0.099 -1.774
-0.593 -0.237 0.503 0.341 -0.501 0.693 -0.519 0.267 -0.687 * 0.044
-1.792 -0.713 1.343 0.977 -1.441 1.362 -1.496 0.768 -1.978
0.085 0.224 0.356 0.154 -0.020 0.195 0.203 0.127 -0.327 0.080
0.453 1.197 1.789 0.784 -0.103 0.688 1.041 0.649 -1.676
0.283 0.242 0.375 -0.039 -0.370 0.485 0.185 0.170 -0.531 * 0.111
1.062 0.905 1.320 -0.139 -1.330 1.201 0.666 0.611 -1.907
-0.030 0.164 -0.007 0.269 -0.293 0.819 ** 0.164 0.148 -0.402 0.089
-0.133 0.729 -0.030 1.143 -1.248 2.409 0.698 0.630 -1.715
Indicesoverall period/sub-
periodsSignificant Holidays
INDICES overall period/sub-periodsMAHA
SHIVARATRI
RAMA
NAVAMIRAMZAN ID
GANESH
CHATURTHI
DUSSERA
MAHANAVAMIDIWALI BAKRID
GURU NANAK
JAYANTHIMOHURRAM
S&P BSE O&GOVERALL PERIOD
(2004-2011)
S&P BSE TECKOVERALL PERIOD
(2001-2011)
S&P BSE AUTOOVERALL PERIOD
(2004-2011)
S&P BSE METALOVERALL PERIOD
(2004-2011)
SUB-PERIOD II
(2002-2011)
S&P BSE PSUOVERALL PERIOD
(2001-2011)
S&P BSE IT
SUB-PERIOD II
(2002-2011)
SUB-PERIOD I
(1999-2001)
OVERALL PERIOD
(1999-2011)
S&P BSE HC
OVERALL PERIOD
(1999-2011)
SUB-PERIOD I
(1999-2001)
372
Table 191: Daily rate of return of BSE and NSE listed indices around important holidays for the overall period (contd..)
Source: Author
Note: * *and * denote Significance at 1 and 5 percent level. Coefficients of Holidays are reported in each cell. Just below the coefficients, the numbers depict t-
statistics.
OTHER DAYS
0.079 -0.015 0.129 0.457 -0.605 * 0.172 -0.171 0.374 -0.514 0.115
0.265 -0.050 0.415 1.484 -1.971 0.388 -0.556 1.218 -1.675
0.213 0.913 0.016 0.614 -0.985 -0.200 -0.314 0.560 -1.255 ** 0.235
0.403 1.665 0.030 1.156 -1.876 -0.262 -0.579 1.066 -2.365
-0.002 -0.489 0.219 0.296 -0.342 0.386 -0.260 0.225 -0.097 0.039
-0.006 -1.387 0.573 0.785 -0.912 0.713 -0.697 0.601 -0.260
-0.449 0.098 0.368 0.539 * -0.314 0.583 0.246 -0.082 -0.394 0.101
-1.826 0.397 1.393 2.079 -1.217 1.546 0.955 -0.316 -1.528
-0.131 0.408 0.218 0.301 -0.727 ** 0.748 0.251 -0.073 -0.448 0.076
-0.456 1.418 0.675 0.957 -2.332 1.684 0.809 -0.236 -1.450
Significant Holidays
INDICES overall period/sub-periodsMAHA
SHIVARATRI
RAMA
NAVAMIRAMZAN ID
GANESH
CHATURTHI
DUSSERA
MAHANAVAMIDIWALI BAKRID
GURU NANAK
JAYANTHIMOHURRAM
CNX INFRAOVERALL PERIOD
(2004-2011)
OVERALL PERIOD
(1999-2011)
BANK NIFTYOVERALL PERIOD
(2000-2011)
CNX IT SUB-PERIOD I
(1999-2001)
SUB-PERIOD II
(2002-2011)
Indicesoverall period/sub-
periods
373
5.4 HALF-MONTH EFFECT:
The Tables 192 to 194 presents the mean daily returns over the first half of the month
for each index and the mean daily return during the rest of the month. According to
Ariel’s (1987), first half of the month is significantly higher when compared to the
second half of the month. The Table-192 presents the significant semi-month and
percentage of positive months having the significant semi-month for broader indices.
We can observe that the first half of the month returns are significantly higher when
compared to second half of the month except in case of BSE Sensex (second sub-period)
and CNX Nifty (for both sub-periods). For the first sub-period, BSE Sensex index has
positive average returns during the first half (0.085 percent) and negative average
returns during the second half of the month (-0.0016). The mean returns during the first
half of the month is higher compared to the second half. The percentage of positive
months is observed to be significantly higher in the first half of the month when
compared to second half of the month. But, in the second sub-period examined the
average returns during first and second half are positive with second half average daily
returns higher than first half average returns. The percentage of positive days are
observed to be significantly higher in the second half of the month. In case of CNX
Nifty index, the average returns as well as the percentage of positive months during the
second half of the month were observed to be significantly higher compared to first half
of the month in both the sub-periods. The CNX Nifty Junior, BSE Mid-cap index, BSE
Small-cap index and NIFTY Mid-cap50 index have positive average returns during the
first half when compared to second half of the month. The percentage of positive month
is also observed to be higher in the first half of the month in all the indices. The Tables
193 and 194 present the results of semi-month effect for all the sectoral indices. All the
BSE and NSE sectoral indices are observed to have positive average returns during both
halves. However, the index returns during the second half of the month are higher
compare to the other half. This is in case of all the BSE listed sectoral indices except for
BSE FMCG index and BSE TECK index have average returns and percentage of
positive months during the second half of the month to be relatively positive and higher
than the first half of the month. Whereas in case of NSE listed sectoral indices, the
results are contrasting to the previous results. Similar to the results obtained for BSE
374
sectoral indices, the CNX IT, Bank Nifty and CNX Infra indices are observed to have
positive average returns during both halves of the calendar month. However, the index
returns during the first half of the month are higher compared to the other half. From
literature, we know that most of the developed country indices seem to hold higher than
average returns during the first half of the month which is similar to the results obtained
in NSE listed indices, mid-cap and small-cap indices, but we also know that, the indices
from the emerging economies, however, show reversed effect which is in accordance
with the results obtained for majority of the indices. Overall, the evidence for the semi-
month anomaly is significant.
5.5 TURN-OF-THE-MONTH EFFECT RESULTS
The Tables 192 to 194 as shown in previous pages also presents the summary of the
turn-of-the-month (Day -1 to Day+3) mean daily returns for all the indices considered
for overall period and sub-periods respectively. The mean daily returns around the turn-
of-the month for BSE and NSE listed broader indices are observed to significantly
higher when compared to other days of the month as shown in Table-192. Considering
BSE Sensex and CNX Nifty index, the daily mean returns were observed to be
significant higher in the second sub-period. The first three days of the months including
the last days of the previous months are found to be significant higher when compared
to other days of the month mainly in the second sub-period examined. Accompanied
with the increasing daily mean returns, the percentage of positive returns is observed to
be significantly higher towards the turn-of-the-month. The percentage of positive returns
is observed to be on average around seventy percentage in case of both the indices. The
percentage can be noticed to be higher in the second sub-period examined when
compared to the first sub-period. Similarly among the broader indices mainly CNX
Nifty Junior, BSE Mid-cap, BSE small-cap and NIFTY Mid-cap50, we can observe the
same trend for the overall period. The mean daily returns around turn-of-the-month are
observed to significantly higher with daily returns increasing from last day of the
previous month (Day -1) until second day of the present month (day +2) after which the
returns decline substantially. Thus turn-of-the-month mean daily returns can definitely
be profitable if buy-hold strategy is followed by the investors. The sum of the mean
375
daily returns for all four days appear to be positive in all the broader indices with
percentage of positive returns substantially supporting the results. Thus on an average,
an investor can make a gain of around two percentage adopting buy-hold strategy only
over the short span of four days towards the turn-of-the-month. The percentage of
positive returns was observed to higher in case of mid-cap and small-cap indices. The
mean daily returns around turn-of-the-month for sectoral indices is presented in Tables
193 and 194 respectively. The results obtained for sectoral indices point towards
normalized returns towards the turn-of-the-month when compared to other days. The
mean daily returns around all the sectoral indices are observed to be high but not
statistically significant when compared to daily returns of other days. The total average
returns earned over four days towards the turn of the month appear to be less when
compared to broader indices ranging from 0.8 percent to 1.8 percent. Whereas, the
percentage of positive returns towards the turn-of-the-month seem to be significantly
higher mainly towards the first and second day of the present month. Thus, when
compared to sectoral indices, broader indices seem to reveal anomaly among daily
returns towards the turn-of-the-month. The sectoral indices have minor or no indication
for turn-of-the-month effect, while broader indices especially mid-cap and small-cap
indices mean daily returns seem to be stronger during the turn-of-the-month. Thus, turn-
of-the-month effect seem to be mostly present in the broader indices.
376
Table-192: Daily stock market returns for BSE and NSE listed broader indices at the turn of the month and semi-month for the overall
period and sub-periods
Source: Author Note: *Significance at 5 percent level
** Significance at 1 percent level
Day -1 Day +1 Day+2 Day+3 Day(-1,+3)
0.2500 0.1640 0.5012 0.1386 1.0538
67.2700 64.9000 68.0800 52.6300 63.2200
0.4558 0.5166 0.3558 0.2087 1.5369
65.96* 74.39** 73.17** 63.41** 69.2325
0.1850 0.1560 0.427* 0.1490 0.9170
60.0000 58.9500 67.0000 51.0000 59.2375
0.4280 0.521** 0.346* 0.2050 1.5000
65.96* 76.8300 72.0000 63.4100 69.5500
0.3240 0.471** 0.369* 0.1460 1.3100
64.1* 68.89** 73.33** 58.8200 66.2850
0.5890 0.6070 0.4410 0.3144 1.9514
67.5* 78.87** 76.71** 71.83** 73.72**
0.5000 0.7550 0.3590 0.2786 1.8926
67.5* 78.87** 78.08** 70.42** 73.72**
0.4700 0.652** 0.542** 0.3450 2.0090
67.57* 76.56** 74** 70.31** 72.1100
First Half (58.28)**
Second Half(56.71)**
Second Half(58.18)**
Second Half (58)**
First Half (57.93)**
Daily stock market returns for BSE and NSE listed indices at the turn of the month and semi-month for the overall period
and sub-period
Mean daily return(%)
positive (%)
overall period/sub-periods Turn of the month
positive (%)
INDICES
Significant semi-month
(percentage of positive
months)
SUB-PERIOD I
(1991-2001)
SUB-PERIOD II
(2002-2011)
SUB-PERIOD I
(1995-2001)
BSE SENSEX
Mean daily return(%)
CNX NIFTY JUNIORMean daily return(%)
positive (%)
Mean daily return(%)
positive (%)
Mean daily return(%)
OVERALL PERIOD
(1997-2011)
SUB-PERIOD II
(2002-2011)
S&P BSE SMALL-CAP
INDEX
Mean daily return(%)
positive (%)
S&P BSE MID-CAP INDEXMean daily return(%)
positive (%)
OVERALL PERIOD
(2005-2011)First Half (63.32)**
OVERALL PERIOD
(2005-2011)First Half (64.54)**
CNX MID-CAP50 INDEXMean daily return(%)
positive (%)
OVERALL PERIOD
(2005-2011)First Half (60.54)**
CNX NIFTY
positive (%)
377
Table-193: Daily stock market returns for BSE and NSE listed indices at the turn of the month and semi-month for the overall period and
sub-periods (contd..)
Source: Author Note: *Significance at 5 percent level
** Significance at 1 percent level
Day -1 Day +1 Day+2 Day+3 Day(-1,+3)
0.4270 0.5500 0.2745 0.4436 1.6951
61.9000 70.75** 65.4200 67.29* 66.3400
0,733 0.4680 0.3060 0.3106 1.0846
62.0000 67.92** 66.3600 65.4200 65.4250
0.4400 0.1500 -0.0840 0.2950 0.8010
65.08* 67.92** 58.8800 62.62* 63.6250
0.4650 0.2440 0.2710 0.3540 1.3340
65.08* 61.3200 71.03** 64.49** 65.4800
0.4310 0.5850 0.2660 0.2060 1.4880
61.9000 64.1500 59.8100 60.7500 61.6525
0.4530 0.6154 0.4440 0.2630 1.7754
68.25** 73** 66.36* 66.36** 68.4925
0.4160 0.4540 0.2750 -0.0130 1.1320
63.7900 65.3100 60.2000 59.1800 62.1200
0.7440 0.3810 0.2360 0.4010 1.7620
76** 66.04* 71.03** 63.5500 69.1550
0.4140 0.4333 0.4360 0.2820 1.5653
65.0000 68.8700 67.0000 64.4900 66.3400
0.6950 0.4960 0.3760 0.2090 1.7760
65.0800 66.0000 64.4900 58.8800 63.6125
OVERALL PERIOD
(1999-2011)Second Half (57.23)**
Daily stock market returns for BSE and NSE listed indices at the turn of the month and semi-month for the overall period
and sub-period (contd..)
INDICES overall period/sub-periods
Significant semi-month
(percentage of positive
months)
Turn of the month
S&P BSE CD INDEX
OVERALL PERIOD
(1999-2011)First Half (55.61)**
S&P BSE CG INDEX
Mean daily return(%)
positive (%)
OVERALL PERIOD
(1999-2011)Second Half (55.62)**
Mean daily return(%)
positive (%)
S&P BSE HCMean daily return(%)
positive (%)
OVERALL PERIOD
(1999-2011)
OVERALL PERIOD
(1999-2011)
Second Half (58.71)**
Second Half (53.89)**
S&P BSE FMCG Mean daily return(%)
positive (%)
S&P BSE PSUMean daily return(%)
positive (%)
OVERALL PERIOD
(2001-2011)
OVERALL PERIOD
(2001-2011)
Second Half (58.51)**
First Half (54.9)**
S&P BSE ITMean daily return(%)
positive (%)
S&P BSE AUTOMean daily return(%)
positive (%)
OVERALL PERIOD
(2004-2011)
OVERALL PERIOD
(2004-2011)
Second Half (57.30)**
Second Half (55.38)**
S&P BSE TECKMean daily return(%)
positive (%)
Mean daily return(%)
S&P BSE O&GMean daily return(%)
positive (%)
OVERALL PERIOD
(2004-2011)Second Half (52.33.38)**
S&P BSE METALpositive (%)
378
Table-194: Daily stock market returns for BSE and NSE listed indices at the turn of the month and semi-month for the overall period and
sub-periods (contd..)
Source: Author Note: *Significance at 5 percent level
** Significance at 1 percent level
Day -1 Day +1 Day+2 Day+3 Day(-1,+3)
0.3240 0.4710 0.3690 0.1460 1.3100
64.1* 68.89** 73.33** 58.5200 66.2100
0.3690 0.6960 0.4610 0.3610 1.8870
61.0200 72.4500 70.4100 61.6200 66.3750
0.5350 0.6040 0.4150 0.2720 1.8260
67.5700 76.5600 71.0000 71.8800 71.7525
OVERALL PERIOD
(1997-2011)First Half (57.93)**
Daily stock market returns for BSE and NSE listed indices at the turn of the month and semi-month for the overall period
and sub-period (contd..)
INDICES overall period/sub-periods
Significant semi-month
(percentage of positive
months)
Turn of the month
BANK NIFTYMean daily return(%)
positive (%)
OVERALL PERIOD
(2000-2011)
OVERALL PERIOD
(2004-2011)
First Half (55.15)**
First Half (58.53)**
CNX IT Mean daily return(%)
positive (%)
CNX INFRAMean daily return(%)
positive (%)