Study
of Stock Market Efficiency and Impact of Crash on Indian and US Market
Falguni
H. Pandya
Ms. Falguni H.Pandya is
a faculty member with Centre for Management Studies, Dharmsinh Desai
University, Nadiad, Gujarat in the Finance Area. Her areas of interest include
Financial Management, Risk Management, Security Analysis and Portfolio
Management and International Financial Management. Her research papers have
been published in national level referred journals and in the form of national
and international conference proceedings.
Email: fhpandya@gmail.com (m) 0 9913879766
Postal
Address
B.No.
45, Gayatri Nagar Society, Nr. G E B, Maktampur Road, Bharuch392012
Hardik
S Shah
Mr. Hardik S Shah is a
faculty member with Centre for Management Studies, Dharmsinh Desai University,
Nadiad, Gujarat in the Finance Area. His areas of interest include Financial
Management, Strategic Finance Management, Taxation, Accounting and Management
of Financial Services. He is pursuing research in the area of Initial Public
Offering.
Email : shah_hardik07@yahoo.in, (m)
8511169690
Postal
Address
Faculty,Centre
for Management Studies (MBA Department)
Dharmsinh
Desai University
College
Road, Nadiad387001, Gujarat, India
Abstract
The study attempts to
find out the market efficiency of Indian and US stock market. For that two main
indices of India namely BSE Sensex and CNX Nifty are studied along with USDow
Jones for a various time brackets between 1993 to 2013. Various tests like KS
Test, Runs test, Auto correlation test, ttest and all descriptive statistics
with JB test are conducted for all three indices for a time bracket of 19932013, 19931995(normal), 19962000 (Before DotCom
Bubble), 20002002 (Dotcom Bubble), 20022007 (Before Credit Crisis), 20072009
(Credit Crisis) and 20092013 (After Credit Crisis). The research summarizes
that Indian stock market is not weak form efficient in all periods however;
from year 2002 onwards stock market exhibits some signs of efficiency.
Key Words:
 Market Efficiency, Indices
Introduction
Emerging markets usually
do not have strict standards in accounting and finance regulations and the
level of market efficiency as that of advanced economies of the world. However,
emerging economy like India and others have physical financial infrastructure
including banks, stock exchange, money market and capital market. In addition,
investment in emerging markets are sought by investors to tap the market
efficiency and thus to earn higher return. However, investments in emerging
markets come with higher risk due to political instability, domestic
infrastructure problems, currency volatility and limited equity opportunities. However,
despite the stage of the different market, it is impossible to ‘beat’ the
market because stock market efficiency causes existing share prices to always
incorporate and reflect all relevant information. In other words, as per
Efficient Market Hypothesis (EMH), stocks always trade at their fair value on
stock exchanges, making it impossible for investors to either purchase
undervalued stocks or sell stocks for inflated prices. Efficient market
hypothesis is statedweak form, semi strong form and strong form of efficiency
on three common forms.
Review
of Literature
Malliaris
and Urrutia (1992) analyzed possible causal relationship among national stock
markets around the October 1987 stock market crash. The paper summarized that
leadlag relationship do not exists before and after market crash. The result
provided empirical evidence of the passive role played by Tokyo but fails to
confirm leading role played by New York or nonJapanese Asian markets during
1987 crash. Schwert (1989) analyzed behavior of stock return
volatility using daily data from 1985 to 1988 and found that on October, 19
there was a largest percentage change in value of the market and stock market
volatility jumped during and after market crash. However, it was also observed
that it returned to normal level more rapidly than any other historical crises.
Research by Yüksel (2002) on Istanbul Stock Exchange during the Russian crisis
in 1998 says that the comparison of the relationship during the crisis period
to those during pre and post crisis period proved that there was a structural
change regarding the price impact of the trading volume. Chancharoenchai and Dibooglu (2006)
examined the volatility spillovers in Southeast Asian emerging markets in the
context of the mid1997 financial crisis, and found spillover effects between
various equity markets. The paper summarized that the sudden fallout in
Thailand seems to have played an important role in the variation in excess
returns in other Southeast Asian markets and it supports the idea of the “Asian
Contagion” suggested that the crisis started in the Thailand and spread over to
other financial markets. Olowe (2009) found that in the Nigerian Stock market
returns show persistence in volatility and clustering and asymmetric
properties.
Kazi et al (2011)
investigated the contagion effect between stock market of US and sixteen OECD
countries due to global financial crisis and it found existence of contagion
effect between US and 16 OECD countries. Boyer
et al (2006) provided empirical evidence that market crises are speeded
globally. Authors categorized the stock markets of emerging economies in to two
parts: those that are eligible for purchase by foreigners and those that are
not. Research proved that greater comovement during high volatility period, especially
for accessible stock index returns indicates that crises spread through the
asset holding of the international investors rather than through changes in
fundamentals. A
novel approach was adopted by Herrero et al (2008) as they applied event study methodology
to analyze whether five south East Asian countries which devalued in 1997
namely Indonesia, Korea, Malaysia, Philippines and Thailand. The study was
based on the assumption that the expectation of devaluation should help the
stock of exporting firms outperform those of nonperforming firms and the study
could find evidence to support this hypothesis, however at different degrees
depending upon the country.
Study
by Salman et al (2012) attempted the short term relationship
between Karachi Stock Exchange, Bombay Stock Exchange, Hong Kong Stock
Exchange, and Tokyo Stock Exchange after the financial crisis of 2007. It
concluded that 1. Tokyo and Karachi Stock Exchange, 2. Hong Kong and Bombay, 3.
Karachi and Bombay (before and after crisis) 4. Tokyo and Bombay have no short
term relationship with each other; while Tokyo and Hong Kong stock exchange
have short term relationship with each other after the financial crisis of
2007. Research by Dhal
(2009) derives some crucial insights from the multivariate co integration
analysis of stock prices indices for the global markets of the US, UK, Japan,
Hong Kong, Singapore and India. Result says that these markets shared single co
integration relationship and the Indian market played a key role and the
analysis showed that global crisis could not have been associated with the
breakdown of the longrun relationships among the markets. Ghosh (2006) tried
to address India’s existence from the effect of financial crisis and analyzed
that India’s success can be attributed to key policy decisions namely
devaluation, IMF, partial liberalization of the domestic financial sector and
graduation opening of the external sector. In a similar track paved by Ghosh (2006);
Study by Jeyanthi et al (2012) summarized that there was no short term as well
as long term negative impact of financial crisis on the Indian stock exchange. Further,
the paper found that the Indian stock market was unaffected by the global
financial crisis. Study by Patel et al (2011) confirmed
the existence of weak form of efficiency in Indian market for a time period of
2004 to 2011. Jethwani, Achuthan (2013) investigated
weak form efficiency of the Indian market during, before and after financial
crisis and the result shows that Indian market is not weak form efficient in
all period and after 2002 it behaves in a more efficient way. This confirms
with the earlier study conducted by Patel et al (2011).
Data Description
Objectives
1.
To
examine the market efficiency level of selected stock markets and to test spillover
effect of the financial crash for a period of 1993 to 2013 for Indian and US
stock market.
2.
To
study and examine the extent of crisis at various periods for the given stock
markets and to derive efficiency level of Indian stock market.
To study the above
mentioned objectives, two main indices of India and one of US are selected as
sample for the study. The daily closing prices of BSE SENSEX, CNX NIFTY 50 and
DOW JONES of last 21 years is collected (1^{st} January 199331^{st}
December 2013). The daily price values of the CNX Nifty 50 and BSE Sensex is collected
from the website of NSE and BSE respectively. Dow Jones data are collected from
yahoo finance. The period was further broken in to 19932013
(Total study period), 19931995 (Normal), 19962000 (Before Dot com Bubble), 20002002
(Dot com Bubble), 20022007 (Before Credit Crisis), 20072009 (Credit Crisis)
and 20092013 (After Credit Crisis). Before conducting any test, it is
required to test whether data are stationary or not and for that Augmented
Dicky Fuller Test (ADF) was conducted.
Returns are calculated
using logarithmic method and it was analyzed by using time series volatility.
Further, to study the variability in stock prices, standard deviation is used
as a proxy. The logarithmic method to calculate return is as follows:
R_{t} = In (I_{t}
/ I_{t1})* 100
Where R_{t} stands for return of
the index and I_{t} indicate index value at time‘t’.
To compute standard deviation following
standard formula was used.
Data collection and
interpretation
Augmented Dicky Fuller
(ADF) test indicates that data are stationary at first difference.
Normality test
Many statistical tests (e.g. ttest)
require that data are normally distributed and therefore it should always be
checked if this assumption is satisfied. The null hypothesis is that the data
is normally distributed and the alternative hypothesis is that the data is not
normally distributed. The KolmogorovSmirnov test is used for sample of more
than 2000. If pvalue is less than 0.05, null hypothesis is rejected.
Here as p value is 0.00
for all three indices (table 1 of appendix A B and C), null hypothesis is
rejected which means that data are not normally distributed. Hence,
nonparametric tests are conducted.
It is a nonparametric
test and is used to test weak form of efficiency of stock market. This test
emphasizes on the direction of change and does not consider the change in
value. The test ignores the type of distribution followed in time series and is
used to check the randomness in time series under consideration. At 5%
confidence level, if the observed value of z is 1.96, null hypothesis is
accepted. Following are the null hypothesis to
check efficiency of the market.
1. Indian
Stock Market is weak form efficient market.
2. Indian Stock market follows random walk
Analysis for Descriptive Statistics:
To study
the above mentioned objectives, one of the basic requirements is that data
should be normal. Usually, main two parameters namely mean and variances are
used to describe the distribution. For that skewness, kurtosis, JarqueBera
test is applied to test the normality of the data. JarqueBera test is a
goodness of fit test of whether the sample data have the skewness and kurtosis
matching a normal distribution. In the data, when the skewness is zero,
kurtosis is three and JB is zero, then it is perfectly normally distributed. Coefficient
of variation shows volatility of the data.
SENSEX
Five (19932013,19931995,
20002002, 20022007, 20072009) out of seven periods under study has negative
skewness, though the value is small; it indicates the tail on the left side is
longer than the right side and the bulk of the values lies including median to
the right side of the mean. Remaining two periods (19962000, 20092013) shows
positive skewness. In case of Kurtosis, four period (19931995, 19962000,
20002002, 20072009) out of seven falls under Platykurtic distribution and other
three periods (19932013, 20022007, 20092013) falls under Leptokurtic
distribution. Therefore, it can be said that distribution of data is not
normal. So on the basis of descriptive statistics, null hypothesis of random
walk is rejected. For Jarque Bera Test, null hypothesis of random walk is
rejected for all periods as value of JB is higher than zero. Coefficient of
Variation in initial two periods (19932013, 19931995) is quite high compare
to other five periods. The value of CV shows that gradually market has become
less volatile. (Refer to Table No.2 of appendix A)
CNX NIFTY
Four (19932013,
20002002, 20022007, 20072009) out of seven periods under study has negative
skewness, and it indicates the tail on the left side is longer than the right
side and the bulk of the values lies including median to the right side of the
mean. Remaining three periods (19931995, 19962000, 20092013) shows positive skewness.
In case of Kurtosis, three period (19931995, 20002002, 20072009) out of
seven falls under Platykurtic distribution and other four periods (19932013, 19962000,
20022007, 20092013) falls under Leptokurtic distribution. Hence, it can be
summarized that distribution of data is not normal. So on the basis of
descriptive statistics, null hypothesis of random walk is rejected. Jarque Bera
test and coefficient of variation indicates same conclusion as that of Sensex. (Refer
to Table No. 2 of Appendix B)
DOWJONES
Five (19932013,19931995,
19962000, 20002002, 20092013) out of seven periods under study has negative
skewness, even if the value is small it indicates the tail on the left side is
longer than the right side and the bulk of the values lies including median to
the right side of the mean. Remaining two periods (20002007, 20072009) are
having positive skewness. In case of Kurtosis, three period (19931995, 20002002,
20022007) out of seven falls under Platykurtic distribution and other four
periods (19932013, 20022007, 20092013) falls under Leptokurtic distribution.
Hence, we can say that distribution of data is not normal. So on the basis of
descriptive statistics, null hypothesis of random walk is rejected. For Jarque
Bera Test, null hypothesis of random walk is rejected for all periods as value
of Jarque Bera is higher than zero. Coefficient of Variation is quite high in
the overall period only. (Refer to Table No. 2 of Appendix C)
In summary, for Sensex except
two periods (19931995, 20072009), remaining five periods are not normally
distributed and which leads to rejection of the null hypothesis of random walk and
similar result is found for CNX Nifty. While for the Dow Jones in all seven periods,
the value of P is significant. That means the data is not normally distributed
which leads to rejection of the null hypothesis of random walk for these
periods.
Runs Test
and analysis
For both the indices
namely Sensex and Nifty, two (20072009, 20092013) out of seven periods the
value of Z is insignificant at 5 % significance level and its value lies inside
the interval of ±1.96, so null hypothesis of random walk is accepted. While in
five (19932013, 19931995, 19962000, 20002002, and 20022007) out of seven
periods, null hypothesis of random walk is rejected (Refer to Table No. 4 of
Appendix A, B). While for the Dow Jones in four periods (19931995, 19962000,
20002002, 20092013) out of seven periods the value of Z is insignificant at 5
% significance level and its value lies inside the interval of ±1.96, so null
hypothesis of random walk is accepted. While in three (19932013, 20022007,
20072009) out of seven periods, null hypothesis of random walk is rejected. (Table
4 of Appendix C).
Autocorrelation
and analysis
Autocorrelation is the
crosscorrelation of a signal with itself as it describes the correlation
between values of the process at different times as a function of the two times
or of the time lag. For null hypothesis to be true, observed serial
correlation should not be statistically significant i.e. should not be greater
than three times the standard error of coefficient. In case of Ljung Box Q statistics, if the value of P <0.05
then it can be said that autocorrelation exist.
For all three indices
namely Sensex, CNX Nifty and Dow Jones in all periods the value of P is
significant which means that autocorrelation exists in the series. In addition
to that, there is at least one lag where correlation coefficient is > 3*Standard
error for all periods. Hence, existence of autocorrelation ultimately rejects
the null hypothesis of random walk in all periods. (Table No. 6 of Appendix A,
B and C).
KruskalWallis
test
KruskalWallis test is
a oneway analysis of variance by ranks and a nonparametric method to test
whether samples originate from the same distribution. The method is used for
comparing more than two samples that are independent or not related and it does
not assume a normal distribution of the residuals. The null hypothesis are.
H_{0}: All Three Stock
market indices follow the random walk for the entire period i.e. from year 1993
to 2013.
H_{0: }Random walk
follows among all three indices before the credit crisis i.e. from year 2002 to
2007.
H_{0: }Random walk
follows among all three indices during the credit crisis i.e. from year 2007 to
2009.
H_{0: }Random walk
follows among all three indices after the credit crisis i.e. from year 2009 to
2013.
Table
A: Kruskal Wallis For The Entire Period
Test
Statistics^{a,b}


RETURN

ChiSquare

1.072

df

2

Asymp. Sig.

.585

a. Kruskal Wallis Test

b. Grouping Variable: INDICES

Table
B: Kruskal Wallis For The Period 20022007
Test
Statistics^{a,b}


RETURN

ChiSquare

26.098

df

2

Asymp. Sig.

.000

a. Kruskal Wallis Test

b. Grouping Variable: INDICES

Table
C: Kruskal Wallis For The Period 20072009
Test
Statistics^{a,b}


RETURN

ChiSquare

.049

df

2

Asymp. Sig.

.976

a. Kruskal Wallis Test

b. Grouping Variable: INDICES

Table
D: Kruskal Wallis For The Period 20092013
Test
Statistics^{a,b}


RETURN

ChiSquare

1.357

df

2

Asymp. Sig.

.507

a. Kruskal Wallis Test

b. Grouping Variable: INDICES

If the P
value is small, it rejects the idea that the difference is due to random
sampling. For the entire period the pvalue is
0.585, for the period of 20022007, the pvalue is 0.000, for the period of 20072009,
the pvalue is 0.976 and for the period of 20092013, the pvalue is 0.507
Consequently, it can be
concluded that only before the period of credit crisis, the difference is not
due to random sampling. Random walk follows between all three indices for
entire period of 21 years as well as during the credit crisis and after the
crisis period. A random walk is defined by the fact
that price changes are independent of each other.
Conclusion
This
study investigates the market efficiency of selected three stock markets. In
addition to that, this research analyses that whether the crisis period alters
the conclusion of efficiency of the stock market. The daily return series of
SENSEX, S&P CNX Nifty and Dow Jones for a period of 1993 to 2013 is considered
for the study. The research summarizes that Indian stock market is not weak
form efficient in all periods however, from year 2002 onwards stock market
exhibits some signs of efficiency. Further, all three independent samples are
compared and is concluded that it follows random walk during and after the
crisis period but not before the crisis period.
References
1. Boyer Brian H., Kumagai Tomomi
and Yuan Kathy (2006), “How Do Crises Spread? Evidence from Accessible and
Inaccessible Stock Indices”, The Journal of Finance, Vol. 61, No. 2 , pp.
9571003
2. Chancharoenchai Kanokwan and
Dibooglu Sel (2006), “Volatility Spillovers and Contagion during the Asian
Crisis: Evidence from Six SouthEast Asian Stock Markets” Emerging Markets
Finance & Trade, Vol. 42, No. 2, pp. 417
3. Dhal Sarat (2009), “ Global
Crisis and the Integration of India's Stock Market”, Journal of Economic
Integration, Vol. 24, No. 4, pp. 778805
4. Ghosh Arunabha (2006), “Pathways
Through Financial Crisis: India”, Global Governance, Vol. 12, No. 4, pp.
413429
5. Herrero Alicia García, Gyntelberg
Jacob and Tesei Andrea (2008), “The Asian crisis: what did local stock markets
expect?” BIS Working Papers No 261
6. Jethwani
Kinjal, Achuthan Sarla (2013), “Stock market efficiency and crisis: evidence
from India” AsiaPacific Finance and Accounting Review ISSN 22781838: Volume
1, No. 2, pp. 3543.
7.
Jeyanthi
B. J. Queensly, William Albert S J, Kalavathy S. Titus (2012), “The Impact Of
Global Financial Crisis On Indian Stock Markets”, International Journal Of
Research In Commerce And Management, Vol. 3, No. 2, pp. 7176.
8. Kazi
Irfan Akbar, Guesmi Khaled and Kaabia Olfa (2011), “Contagion
Effect of Financial Crisis on OECD Stock Markets”, Discussion Paper No.
201115. Accessed from http://www.economicsejournal.org/economics/discussionpapers/201115
as on 12122013.
9. Malliaris A. G. and Urrutia Jorge
L. (1992), “The International Crash of October 1987: Causality Tests”, The
Journal of Financial and Quantitative Analysis, Vol. 27, No. 3, pp.353364
10. Olowe Rufus Ayodeji (2009),
“Stock Return Volatility, Global Financial Crisis And The Monthly
Seasonaleffect On The Nigerian Stock Exchange” African Review of Money Finance
and Banking, pp. 73107
11. Patel
Nikunj R., Patel Bhavesh K and Ranpura Darshan (2011), “Testing Weak Form
Market Efficiency Of Indian Stock Markets”, Ss International
Journal Of Business And Management Research, Vol. 1, No. 3, pp. 128.
12. Salman
Nabeel, Abbas Muhammad, Ahmad Mushtaq, and Majid Muhammad Abdul, Khan Noheed
and Raheel Awad (2012), “Impact of financial crisis on Asian stock Markets”,
interdisciplinary journal of contemporary research in business Vol 3, No 11.
pp. 736741.
13. Schwert G. William (1989), “Stock
Volatility and the Crash of '87”, The Review of Financial Studies, Vol. 3, No.
1, ,pp. 77102
14.
Yüksel Aydin
(2002), “The Performance of the Istanbul Stock Exchange during the Russian
Crisis”, Emerging Markets Finance & Trade, Vol. 38, No. 6, , pp. 7899
Websites
ü en.wikipedia.org
ü www.investopedia.com
ü www.moneycontrol.com
ü www.nseindia.com
ü www.bseindia.com
ü finance.yahoo.com
ü money.cnn.com
ü www.google.com
ü www.slideshare.net
Appendices
Table A:ADF for BSE
ADF
Test Statistic

33.55698

1% Critical Value*

3.4348



5% Critical Value

2.8626



10% Critical Value

2.5674

*MacKinnon
critical values for rejection of hypothesis of a unit root.

Augmented
DickeyFuller Test Equation

Dependent
Variable: D(CLOSE,2)

Method:
Least Squares

Date:
12/19/13 Time: 16:44

Sample(adjusted):
1/11/1993 5/04/2012

Included
observations: 5040 after adjusting endpoints

Variable

Coefficient

Std. Error

tStatistic

Prob.

D(CLOSE(1))

1.009154

0.030073

33.55698

0.0000

D(CLOSE(1),2)

0.082824

0.026836

3.086319

0.0020

D(CLOSE(2),2)

0.078042

0.023238

3.358451

0.0008

D(CLOSE(3),2)

0.054351

0.019199

2.830950

0.0047

D(CLOSE(4),2)

0.046042

0.014088

3.268078

0.0011

C

3.389471

2.206431

1.536179

0.1246

Rsquared

0.464436

Mean dependent var

0.008262

Adjusted
Rsquared

0.463904

S.D. dependent var

213.7032

S.E.
of regression

156.4704

Akaike info criterion

12.94480

Sum
squared resid

1.23E+08

Schwarz criterion

12.95257

Log
likelihood

32614.90

Fstatistic

873.0882

DurbinWatson
stat

2.004013

Prob(Fstatistic)

0.000000

Table B: ADF for NSE
ADF
Test Statistic

33.34988

1% Critical Value*

3.4348



5% Critical Value

2.8626



10% Critical Value

2.5674

*MacKinnon
critical values for rejection of hypothesis of a unit root.

Augmented
DickeyFuller Test Equation

Dependent
Variable: D(CLOSE,2)

Method:
Least Squares

Date:
12/19/13 Time: 16:50

Sample(adjusted):
1/11/1993 6/08/2012

Included
observations: 5065 after adjusting endpoints

Variable

Coefficient

Std. Error

tStatistic

Prob.

D(CLOSE(1))

1.004996

0.030135

33.34988

0.0000

D(CLOSE(1),2)

0.067270

0.026919

2.499027

0.0125

D(CLOSE(2),2)

0.069741

0.023318

2.990804

0.0028

D(CLOSE(3),2)

0.047503

0.019269

2.465217

0.0137

D(CLOSE(4),2)

0.041637

0.014057

2.962037

0.0031

C

0.998543

0.663774

1.504342

0.1326

Rsquared

0.469785

Mean dependent var

0.004859

Adjusted
Rsquared

0.469261

S.D. dependent var

64.77475

S.E.
of regression

47.18961

Akaike info criterion

10.54741

Sum
squared resid

11265683

Schwarz criterion

10.55514

Log
likelihood

26705.31

Fstatistic

896.4813

DurbinWatson
stat

2.003658

Prob(Fstatistic)

0.000000

Appendix A BSE Sensex
Table
1:Test for Normality:
Tests
of Normality


KolmogorovSmirnov^{a}


Statistic

df

Sig.

BSE

.051

5147

.000

a. Lilliefors Significance Correction

Table
2:Descriptive statistics of different period
N

TIME PERIOD

MEAN

STD. DEV.

COE. OF VAR.

SKEWNESS

KURTOSIS

JORQUE BERRA

5045

19932013

0.040

1.638

40.689

0.099

4.967

5194.549

676

19931995

0.030

1.533

51.099

0.012

1.560

68.574

1025

19962000

0.052

1.767

33.964

0.032

1.869

149.383

645

20002002

0.091

1.725

19.050

0.410

2.139

141.015

1272

20022007

0.150

1.398

9.339

0.768

6.650

2468.519

346

20072009

0.207

2.677

12.960

0.149

1.037

16.764

1243

20092013

0.063

1.380

21.828

1.152

15.550

12798.870

Table
3:Descriptive statistics of each year
N

TIME

MEAN

STD. DEV.

CO. OF VAR.

SKEWNESS

KURTOSIS

JORQUE BERRA

214

1993

0.129

1.864

14.467

0.442

1.153

18.820

231

1994

0.069

1.435

20.716

0.577

1.783

43.419

231

1995

0.101

1.261

12.502

0.115

0.385

1.934

238

1996

0.003

1.522

443.773

0.504

0.974

19.508

246

1997

0.069

1.638

23.622

0.309

4.315

194.741

244

1998

0.074

1.908

25.821

0.046

1.306

17.413

248

1999

0.199

1.815

9.118

0.053

1.492

23.121

250

2000

0.093

2.204

23.822

0.243

0.946

11.787

248

2001

0.079

1.719

21.655

0.462

1.679

37.976

251

2002

0.014

1.102

79.903

0.142

1.494

24.175

254

2003

0.216

1.166

5.410

0.190

0.133

1.712

254

2004

0.048

1.610

33.266

1.554

13.748

2102.653

251

2005

0.141

1.080

7.682

0.442

0.405

9.890

250

2006

0.153

1.627

10.616

0.485

3.002

103.683

249

2007

0.155

1.544

9.954

0.210

1.455

23.786

246

2008

0.302

2.853

9.442

0.069

0.904

8.567

243

2009

0.244

2.186

8.950

1.313

10.922

1277.570

252

2010

0.064

1.006

15.782

0.254

0.650

7.149

247

2011

0.115

1.321

11.529

0.279

0.058

3.241

251

2012

0.091

0.919

10.082

0.088

0.623

4.389

250

2013

0.034

1.096

31.865

0.111

1.429

21.779

Table
4:Runs test
TIME PERIOD

TEST VALUE a

CASES<TEST VALUE

CASES>=TEST VALUE

TOTAL CASES

NUMBER OF RUNS

Z

P VALUE









19932013

0.0402

2454

2591

5045

2299

6.274

0

19931995

0.03

339

336

675

270

5.277

0

19962000

0.052

521

504

1025

480

2.086

0.037

20002002

0.0905

306

339

645

289

2.659

0.008

20022007

0.1496

602

670

1272

593

2.373

0.018

20072009

0.2065

173

173

346

160

1.507

0.132

20092013

0.07

621

622

1243

599

1.334

0.182

Table
5:T Test
PERIOD

T

DF

SIGNIFICANCE

MEAN DIFF

LOWER

UPPER

19932013

1.746

5044

0.081

0.040

0.005

0.085

19931995

0.509

674

0.611

0.030

0.086

0.146

19962000

0.943

1024

0.346

0.052

0.056

0.160

20002002

1.333

644

0.183

0.091

0.224

0.043

20022007

3.819

1271

0.000

0.150

0.073

0.227

20072009

1.435

345

0.152

0.207

0.490

0.077

20092013

1.615

1242

0.107

0.063

0.014

0.140

Table
6:Auto correlation test
PERIOD

LAGS

1

2

3

4

5

6

7

8

9

10

19932013

AC

0.434

0.077

0.003

0.024

0.003

0.046

0.023

0.009

0.003

0.024


STD. ERROR

0.014

0.014

0.014

0.014

0.014

0.014

0.014

0.014

0.014

0.014


Q Stat

951.428

981.670

981.722

984.716

984.757

995.237

997.999

998.447

998.495

1001.000


Prob

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

19931995

AC

0.276

0.224

0.031

0.007

0.064

0.024

0.032

0.053

0.017

0.098


STD. ERROR

0.038

0.038

0.038

0.038

0.038

0.038

0.038

0.038

0.038

0.038


Q Stat

51.485

85.393

86.055

86.087

88.840

89.219

89.899

91.836

92.024

98.604


Prob

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

19962000

AC

0.485

0.034

0.023

0.036

0.021

0.076

0.064

0.021

0.025

0.020


STD. ERROR

0.031

0.031

0.031

0.031

0.031

0.031

0.031

0.031

0.031

0.031


Q Stat

242.010

243.201

243.724

245.087

245.529

251.549

255.763

256.199

256.844

257.273


Prob

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

20002002

AC

0.419

0.066

0.053

0.063

0.014

0.026

0.005

0.017

0.045

0.025


STD. ERROR

0.039

0.039

0.039

0.039

0.039

0.039

0.039

0.039

0.039

0.039


Q Stat

113.744

116.556

118.387

120.937

121.061

121.508

121.525

121.710

123.009

123.429


Prob

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

20022007

AC

0.420

0.139

0.031

0.067

0.009

0.039

0.009

0.010

0.010

0.073


STD. ERROR

0.028

0.028

0.028

0.028

0.028

0.028

0.028

0.028

0.028

0.028


Q Stat

224.359

248.858

250.069

255.753

255.853

257.822

257.924

258.065

258.181

265.090


Prob

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

20072009

AC

0.445

0.055

0.060

0.104

0.037

0.036

0.016

0.152

0.082

0.032


STD. ERROR

0.054

0.054

0.053

0.053

0.053

0.053

0.053

0.053

0.053

0.053


Q Stat

68.893

69.946

71.190

74.952

75.440

75.907

75.993

84.186

86.575

86.939


Prob

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

20092013

AC

0.455

0.037

0.025

0.037

0.021

0.035

0.046

0.033

0.047

0.004


STD. ERROR

0.028

0.028

0.028

0.028

0.028

0.028

0.028

0.028

0.028

0.028


Q Stat

257.482

259.186

259.976

261.651

262.191

263.700

266.295

267.692

270.476

270.499


Prob

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

Appendix B: CNX Nifty NSE
Table
1: Test for Normality
Tests
of Normality


KolmogorovSmirnov^{a}


Statistic

df

Sig.

NSE

.054

5172

.000

a. Lilliefors Significance Correction

Table 2:
Descriptive statistics of different periods
N

TIME PERIOD

MEAN

STD. DEV.

COF. OF VAR.

SKEWNESS

KURTOSIS

JORQUE BERRA

5070

19932013

0.040

1.624

40.323

0.129

6.097

7865.758

679

19931995

0.029

1.469

50.243

0.043

1.078

33.095

1047

19962000

0.054

1.761

32.488

0.161

3.207

453.170

645

20002002

0.080

1.606

19.995

0.353

2.565

190.204

1272

20022007

0.143

1.464

10.227

0.962

7.728

3361.353

346

20072009

0.193

2.619

13.534

0.345

1.785

52.800

1243

20092013

0.061

1.381

22.702

1.210

16.200

13895.013

Table
3:
Descriptive statistics of each year
N

PERIOD

MEAN

STD. DEV.

COF. OF VAR.

SKEWNESS

KURTOSIS

JORQUE BERRA

213

1993

0.158

1.744

11.046

0.347

0.365

5.457

230

1994

0.055

1.400

25.612

0.634

1.961

52.260

236

1995

0.112

1.242

11.130

0.106

0.752

5.996

250

1996

0.004

1.526

365.632

0.696

1.103

32.820

244

1997

0.075

1.798

24.010

0.058

7.564

581.841

250

1998

0.080

1.777

22.280

0.094

1.617

27.594

254

1999

0.203

1.837

9.056

0.045

2.247

53.510

250

2000

0.063

2.002

31.591

0.105

1.491

23.620

248

2001

0.071

1.630

22.900

0.462

2.262

61.672

251

2002

0.013

1.061

83.184

0.078

1.457

22.463

254

2003

0.213

1.232

5.777

0.337

0.470

7.134

254

2004

0.040

1.763

44.122

1.802

14.397

2331.090

251

2005

0.124

1.114

9.017

0.517

0.592

14.831

250

2006

0.134

1.650

12.305

0.620

2.731

93.720

249

2007

0.175

1.601

9.130

0.258

1.558

27.955

246

2008

0.297

2.808

9.467

0.283

1.688

32.505

243

2009

0.232

2.143

9.233

1.508

12.621

1704.950

252

2010

0.066

1.024

15.634

0.277

0.670

7.939

247

2011

0.114

1.321

11.548

0.270

0.057

3.045

251

2012

0.097

0.955

9.801

0.076

0.662

4.818

250

2013

0.026

1.138

43.529

0.116

1.514

24.447

Table
4:
Runs test
TIME PERIOD

TEST VALUE a

CASES<TEST VALUE

CASES>=TEST VALUE

TOTAL CASES

NUMBER OF RUNS

Z

P VALUE









19932013

0.0403

2483

2587

5070

2293

6.799

0

19931995

0.0293

346

332

678

261

6.064

0

19962000

0.0542

537

510

1047

469

3.413

0.001

20002002

0.0803

308

337

645

283

3.147

0.002

20022007

0.1432

593

679

1272

591

2.429

0.015

20072009

0.1935

164

182

346

168

0.597

0.55

20092013

0.05

621

622

1243

607

0.88

0.379

Table
5:
Ttest
PERIOD

T

DF

SIDNIFICANCE

MEAN DIFF

LOWER

UPPER

19932013

1.766

5069

0.077

0.040

0.004

0.085

19931995

0.519

677

0.604

0.029

0.082

0.140

19962000

0.996

1046

0.319

0.054

0.053

0.161

20002002

1.27

644

0.204

0.080

0.205

0.044

20022007

3.487

1271

0.001

0.143

0.063

0.224

20072009

1.374

345

0.170

0.193

0.470

0.083

20092013

1.553

1242

0.121

0.061

0.016

0.138

Table
6:
Auto correlation test
PERIOD

LAGS

1

2

3

4

5

6

7

8

9

10

19932013

AC

0.433

0.084

0.009

0.018

0.013

0.053

0.024

0.005

0.004

0.042


STD. ERROR

0.014

0.014

0.014

0.014

0.014

0.014

0.014

0.014

0.014

0.014


Q Stat

952.337

988.317

988.750

990.450

991.253

1005.383

1008.201

1008.328

1008.416

1017.488


Prob

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

19931995

AC

0.242

0.268

0.004

0.006

0.018

0.003

0.057

0.099

0.019

0.100


STD. ERROR

0.038

0.038

0.038

0.038

0.038

0.038

0.038

0.038

0.038

0.038


Q Stat

39.869

88.876

88.885

88.910

89.139

89.146

91.396

98.057

98.296

105.211


Prob

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

19962000

AC

0.491

0.030

0.026

0.003

0.039

0.086

0.046

0.025

0.002

0.071


STD. ERROR

0.031

0.031

0.031

0.031

0.031

0.031

0.031

0.031

0.031

0.031


Q Stat

253.300

254.300

254.990

255.000

256.600

264.460

266.640

267.300

267.300

272064.000


Prob

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

20002002

AC

0.387

0.120

0.030

0.043

0.035

0.036

0.046

0.056

0.020

0.014


STD. ERROR

0.039

0.039

0.039

0.039

0.039

0.039

0.039

0.039

0.039

0.039


Q Stat

96.830

106.900

107.510

108.710

109.510

110.360

111.770

113.820

114.080

114.220


Prob

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

20022007

AC

0.406

0.150

0.037

0.069

0.017

0.028

0.012

0.011

0.002

0.089


STD. ERROR

0.028

0.028

0.028

0.028

0.028

0.028

0.028

0.028

0.028

0.028


Q Stat

210.073

241.870

243.576

249.644

250.013

250.984

251.178

251.343

251.351

261.438


Prob

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

20072009

AC

0.461

0.030

0.044

0.086

0.042

0.062

0.007

0.140

0.070

0.046


STD. ERROR

0.054

0.054

0.053

0.053

0.053

0.053

0.053

0.053

0.053

0.053


Q Stat

73.867

74.260

74.936

77.515

78.135

79.491

79.510

86.436

88.182

88.945


Prob

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

20092013

AC

0.469

0.020

0.027

0.039

0.021

0.045

0.058

0.039

0.047

0.001


STD. ERROR

0.028

0.028

0.028

0.028

0.028

0.028

0.028

0.028

0.028

0.028


Q Stat

273.528

274.037

274.950

276.821

277.361

279.904

284.134

286.079

288.879

288.880


Prob

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

Appendix C: Dow Jones
Table
1: Test for Normality
Tests
of Normality


KolmogorovSmirnov^{a}


Statistic

df

Sig.

DOW JONES

.080

5288

.000

a. Lilliefors Significance Correction

Table
2:
Descriptive statistics of different periods
N

TIME PERIOD

MEAN

STD. DEV.

CV

SKEW

KURT

JB

5288

19932013

0.030

1.128

37.019

0.165

8.240

14985.608

756

19931995

0.058

0.598

10.370

0.361

1.758

113.743

1059

19962000

0.063

1.092

17.448

0.574

4.292

871.055

647

20002002

0.048

1.403

29.325

0.076

2.611

184.335

1274

20022007

0.051

0.835

16.411

0.216

2.717

401.955

355

20072009

0.170

2.269

13.318

0.262

3.526

187.931

1197

20092013

0.065

0.999

15.355

0.355

3.186

531.374

Table
3:
Descriptive statistics of each year
N

YEAR

MEAN

STD. DEV.

CV

SKEW

KURT

JB

252

1993

0.050

0.548

10.940

0.379

2.250

59.160

252

1994

0.008

0.688

81.942

0.331

1.324

22.993

252

1995

0.115

0.544

4.752

0.229

1.318

20.439

254

1996

0.091

0.755

8.298

0.593

1.776

48.264

253

1997

0.081

1.184

14.676

0.846

6.583

487.012

252

1998

0.059

1.256

21.203

0.548

4.274

204.394

252

1999

0.089

1.017

11.396

0.044

0.100

0.187

252

2000

0.025

1.309

51.813

0.283

1.738

35.093

248

2001

0.030

1.350

45.447

0.571

4.065

184.215

252

2002

0.073

1.604

22.023

0.494

1.234

26.229

252

2003

0.090

1.043

11.646

0.112

1.133

14.021

252

2004

0.012

0.683

55.536

0.010

0.115

0.143

252

2005

0.002

0.649

268.501

0.004

0.040

0.018

251

2006

0.060

0.622

10.338

0.110

1.239

16.550

251

2007

0.025

0.918

36.960

0.623

1.637

44.268

253

2008

0.163

2.381

14.583

0.227

3.822

156.163

252

2009

0.068

1.524

22.274

0.072

2.181

50.180

252

2010

0.041

1.018

24.541

0.177

2.162

50.415

252

2011

0.021

1.328

62.201

0.532

2.572

81.332

250

2012

0.028

0.743

26.513

0.027

0.916

8.774

252

2013

0.093

0.640

6.860

0.202

1.362

21.176

Table
4:
Runs test
TIME PERIOD

TEST VALUE a

CASES<TEST VALUE

CASES>=TEST VALUE

TOTAL CASES

NUMBER OF RUNS

Z

P VALUE









19932013

0.05

2644

2644

5288

2758

3.108

0.002

19931995

0.07

378

378

756

386

0.51

0.61

19962000

0.07

529

530

1059

520

0.646

0.519

20002002

0.05

323

324

647

332

0.59

0.555

20022007

0.06

637

637

1274

691

2.971

0.003

20072009

0.1

177

178

355

200

2.286

0.022

20092013

0.07

598

599

1197

619

1.128

0.259

Table
5:
T test
PERIOD

T

DF

SIGNIFICANCE

MEAN DIFF

LOWER

UPPER

19932013

1.964

5287

0.050

0.030

0.000

0.061

19931995

2.651

755

0.008

0.058

0.015

0.100

19962000

1.865

1058

0.062

0.063

0.003

0.128

20002002

0.867

646

0.386

0.048

0.156

0.060

20022007

2.175

1273

0.030

0.051

0.005

0.097

20072009

1.415

354

0.158

0.170

0.407

0.067

20092013

2.253

1196

0.024

0.065

0.008

0.122

Table
6:
AUTO CORRELATION TEST
PERIOD

LAGS

1

2

3

4

5

6

7

8

9

10













19932013

AC

0.507

0.023

0.040

0.011

0.047

0.043

0.040

0.032

0.027

0.050


STD. ERROR

0.014

0.014

0.014

0.014

0.014

0.014

0.014

0.014

0.014

0.014


Q Stat

1360.000

1360.000

1370.000

1370.000

1380.000

1390.000

1400.000

1410.000

1410.000

1420.000


Prob

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

19931995

AC

0.485

0.022

0.039

0.017

0.028

0.019

0.015

0.004

0.024

0.020


STD. ERROR

0.036

0.036

0.036

0.036

0.036

0.036

0.036

0.036

0.036

0.036


Q Stat

178.266

178.632

179.794

180.013

180.605

180.877

181.057

181.067

181.520

181.830


Prob

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

19962000

AC

0.473

0.014

0.047

0.052

0.033

0.037

0.037

0.008

0.000

0.107


STD. ERROR

0.031

0.031

0.031

0.031

0.031

0.031

0.031

0.031

0.031

0.031


Q Stat

237.207

237.417

239.815

242.693

243.885

245.330

246.798

246.868

246.868

259.059


Prob

