CHAPTER # 1

1. INTRODUCTION

1.1. Definition of herding

There are various definitions of herding, according to Nofsinger and Sias, (1999),

herding is a that group of investors that use to trade in similar direction over a specific

time period. Another study reveals that herd behavior is occur frequently when

various people take similar actions because some are definitely copying the behavior

of other without thinking about its advantages and disadvantages (Graham, 1999).

It is a kind of behavior that many people take same action and correlation across the

individuals is high (Devenow and Welch, 1996).From the definition use for our study

we are agree that investor that take similar path in their investment behavior will be

considered as herding behavior.

Khoshsirat & salari (2011), suggests that when investors follow market blindly they

pay no heed to their personal information consequently the variation between

individual stock return and the returns of the market become less.

When unaware investors go behind the market, it can cause volatility (Wang,

1993) .this type of investors pay money when there is upward movement of the

market and put up for sale at the time of downward movement ( Avery &

Zemsky,1998).

Following each other might change the circumstances and blind (foolish) investor

mimic the blind one. This situation often creates bubbles. It become clear that detailed

1
information is not involved in gaining unusual returns (Park, 2010). Before crash of

the market in March 2008, Karachi Stock Exchange (KSE) faced this situation (Smith,

2011).

Kodres and Pritsker (1997) find herd behavior through banks, dealers, brokers and

hedge funds.

Generally managers are not sure that either those investors are taking right decision or

not whom they are following ( Banerjee, 1992).this type of dealing when investor’s

decision is related to another would limit the financial market and would weaken the

whole market system (Bikhchandani and Sharma, 2001; Chari and Kehoe,2004).

In market plenty of information is neglected on a daily basis. At the time of making

decision an investor always think that which information should I use? Ordinarily,

they start thinking about other traders; they study their actions and start investing by

observing their actions rather than on the basis of their own judgments (Ahsan &

Sarkar, 2013).

Proper diversification is very necessary for investment. Sometime investors imagine

high profit and take avoidable risk. Due to this greediness, investors follow the other

financial experts blindly (Malik & Elahi, 2014).

Diversity of ideas and beliefs of investors make it very complex and essential to bring

together those viewpoints (Belhoula & Naoui, 2011). Many researchers uses Cross

sectional absolute deviation (CSAD) and cross sectional standard deviation (CSSD) to

measure herd behavior and regression is run for the values of CSSD and CSAD .

When the values of the two ‘Betas’ turn out to be negative, herd behavior will be

present in that particular market (Javed, Zafar, & Hafeez, 2009).

2
Chiang, Mason, Tan,& Nelling (2008), study herd behavior in China market By using

the data from 1996-2003. The results of their study shoe the evidence of herding.

The investor who gathers information and does efforts is caller leader investor the

later investors imitate them not knowing that either they are taking correct choice

(Khanna & Mathew, 2011).

1.2. Extreme market movement and financial turmoil

(Kaminsky and Schmukler, 1999) inspect the financial calamity in Asia and argue the

damaging outcomes of news and discussed the time of market movement (upward &

downward), the existence of imitating behavior can cause severe damages and worse

financial condition.

Another study explains chances and reasons of crashes and bubbles in the market and

discuss that short-term investments shift the asset prices from their original worth

(Patterson and Sharma, 2007).

If there constant herding behavior exists in the market it would extremely increase or

decrease the capital (Kumar and Prasad, 2002).

1.3. Research Objective

The main purpose of the study is to examine whether herd behavior exists in

Pakistan’s stock market.

1.4. Problem statement

As the notion of herd behavior is checked in Pakistan’s stock market, therefore, the

problem statement will be formulated as;

3
 Does herd behavior exist in Pakistani stock market and how it affects the stock

return?

1.5. Arrangement of the thesis

The remaining portion of the thesis is structured as stated below;

Chapter 2 comprises literature review (previous studies) and ideas onherding behavior

in different stock markets. Chapter 3 includesdata source and the methodology. After

this chapter observed results and the extent to which the herd behavior exists in

Pakistan’s stock market is stated in chapter 4. In the end, part 5 provides summary of

the thesis then conclusion and limitation of the research.

4
 CHAPTER # 2

2. LITERATURE REVIEW

In this chapter the concept of herding is explained. There are many concepts relating

to herding and herding behavior of investors in stock exchanges. Numerous previous

researchers have contributed to explain the different concepts of herding behavior.

The first part of this chapter defines the basic concept of herd behavior, different

types of herding and then focuses on herding behavior of investors in stock

exchanges. Basically this perception is related to behavioral finance. After clearing

the concept of herding we find the vital basic reasons behind herding. Measurement of

herding is also given meticulous attention. In the end we have description of this

chapter.

Sometimes investors preplan to imitate the actions, deeds and procedures of other

investors. There are several views about the reasons that compel the investor to

restrain its personal signals and go behind the crowed. Those reasons can be profit or

greediness. Similarity in investment choices amongst investors is actually herd

behavior in finance. Usually small investors imitate the well informed and

experienced investors and call them ‘financial gurus’ with the thought that their own

information is not good enough to invest (Bikhchandani and Sharma, 2000).

The investors, who are sound educated and rational, try to avoid the notion of herding.

Indian investors get the information of stock exchanges by using technology that is

why herding behavior does not exist in Indian stock markets (Prosad, Kapoor and

Sengupta, 2012).

5
2.1. Individual verses institutional investors

There are two sort of investor, first one is individual investor, in comparison with

institutional investors, have less information and are more exposed to behavioral

biases and market reactions (kaniel et al, 2008; Barber and Odean, 2008). As

individual investors have lesser information, they sensibly act as price chaser (Wang

1993, 1994). For that reason, these investors are said to be more responsive to market

return fluctuation. Sometimes individual investors mimic other investors because of

the irrational and systematic response to the fashion or trend (Nofsinger and Sias,

1999).

Second one is institutional investors, there are various reasons due to which they

engage in herding, and these may be, agency problem, safety character and trend.

Though, there is no proper study to examine the difference between the two sorts of

investors on the basis of their herd behavior.

Nofsinger, (1996) examined the degree to which institutions and individual investors

engage in herding. The general finding is that institutional investors herd to a greater

extent than individual investor. Moreover, institutional herding is concentrated during

specific market conditions such as high uncertainty in the direction of the market.

Herding is most likely during periods of high uncertainty in the market and in stocks

with high volume.

As far as herding in institutional investors is concern, it comes out from both rational

and irrational type of investor behavior. Welch (2000), explain that when investor pay

no heed to their past beliefs and go behind the trial of other investors, this is called

irrational type of study or observation. Conversely, in rational observation, principle-

6
agent dilemma is involved in which administrator copy the actions. Hence, to

maintain the reputation in the market they absolutely close the eyes to their personal

information (froot et al.1992; Scharfstein and Stein 1990).

2.2. Reasons behind herding

Some investors plan a decision but when they observe other investors gaining profit

or better feedback they reverse their decision. There are some reasons of this sort of

behavior;

1. The other investor’s actions show that those better knew the investment return.

2. Usually the managers who act as an agent, and invest on behalf of

shareholders they get incentives on good performance. Imitative behavior can

get them better compensation.

3. Another reason of following others is that investors have their own preference.

(Bikhchanani & Sharma, 2001)

Singhvi (2001), conclude that human beings are sensitive towards the following six

factors;

 Achievements (success stories) of the other investors.

 When media concentrate on the particular stock market.

 Get rich quick philosophy

 To earn money easily without doing much work.

 Internet stock’s performance.

 Greediness of the investors.

7
2.3. Compensation based herding

As it is proposed that money manager act as an agent, he is often worried about the

rewards and compensation, but if its compensation depends on his performance in

comparison with other professional mangers. This could alter the agent’s incentives

and sometimes they makes inefficient or disorganized portfolio (Brennan, 1993; Roll,

1992). In order to avoid this danger, mostly managers copy the expert managers or

professionals.

Maug & Naik, (1996) explains this idea by the concept of agent. Managers are not

risk taker they avoid risk and their compensation is based on performance. The

investor and the agent both have their own personal information about the returns.

When an investor takes the decision on the basis of his information the agent then

take investment decision by following that investor. Due to this agent’s portfolio go in

line with that investor’s portfolio as he copied its actions.

2.4. Informational cascades

Bikhchandani & Sharma, (2001) examine that an investor observes that the number of

investors who invested is more than the number of investors who did not invest; they

start investment and pay no attention to the personal information. They herd on the

basis of other’s decision which is absolutely wrong for all investors.

A study on Asian crises suggests that this behavior is a sort of infection or

contamination. In all markets the correlation coefficients are supposed as

representative of herding behavior. The author explains when the disaster becomes

open to the communitythe investors choice meet due to herding behavior (Chiang et

al, 2007).

8
Lin and Swanson (2003) investigate herding behavior Taiwan stock exchange from

the year 1996-2003, but their centre of attention was foreign investors. They realize

that foreign investors are not involved in herding as they observe the results by the

methodology of change et al (2000).

2.5. Measurement of herd behavior

Numerous methods of measuring herd behavior in stock exchanges have been

developed to empirically examine herd behavior in market. These methods are figure-

based. There are various types of herding therefore it is necessary to differentiate

them e.g. intentional and spurious herding. It is not easy to declare the exact

fundamentals of this behavior that is why it is difficult to measure and calculate it

(Bikhchandani & Sharma, 2001).

For this purpose dispersion between the stock return is checked and for the

measurement of dispersion cross sectional standard deviation is used. During the

period of market stress the decrease in dispersion points to presence of herding

behavior among investors. Christie and Huang (1995) examine US market and

conclude that during market stress period mostly investors are expected to herd.

Another similar but less accurate method of detecting herd behavior is Absolute Cross

Sectional deviation (CSAD), developed by Chang et al (2000). They suggested South

Korea and Taiwan markets are involve in herding by using this method.

2.6. Previous studies

Herding behavior was tested in Italian Stock Market by Caparrelli et al, (2004). They

used the method of Cheng et al (2000) and found that herd pattern is non linear. They

9
find out herd behavior by using H-statistics proposed by Hwang and Salmon (2001) to

distinguish intentional and spurious herding. The decrease in H- statistic is point to

intentional herding. It is greater in upward market movement.

For measuring herd behavior the methodology of Christie and Huang (1995) is

broadly used. Numeral other studies have applied this methodology.

Empirical study is also performed in Australian stock markets. In this study intra-day

and daily herding was observed, but in that particular market there was no indication

of herd behavior (Henker et al, 2006). Another study focuses Athens stock market

from the time period 1985 to 2005. In that stock market too, there was no indication

of herd behavior (Tessaromatis and Thomas, 2009). Hwang and salmon (2004)

anticipated another measurement. Their methodology is quite different from the other

methodology of proceeding studies, (e.g. Chiang and Zheng, 2010; chang et al, 2000)

which conclude that herding is present is US equity market.

Behavioral study is not only performed in Asian market but herd behavior is also

observed in four European markets using the methodology by (Hwang and Salmon,

2004). In France, Germany, Italy and UK data supports the presence of herding

behavior (Khan et al, 2011). Fascinatingly, during market crises and turmoil

researcher find no indication of herd behavior. Chiang et al (2007) conducted study on

Asian crises and focus on an aspect of behavioral finance, herding behavior and his

find that high correlation between the stock returns are found and it is said to have

herding behavior of investors.

Khanna and Mathew (2011), suggests that information is often misplaced when

decision makers herd investors make decision in future cannot deduce any

confidential information, collected by party that make a decision to herd . Herding

10
may also contribute to poor decision making and benefit of society loss. Society may

benefit from superior decision and better information aggregation.

If we study the forces that cause herding behavior and at the time of investment an

investor pay no heed to his personal information, according to Scharftein and stein,

(1988), this behavior is unproductive from social point of view but it is logical from

the viewpoint of an investor if he is anxious about the reputation.

Cui (1992) provides the extension of Bikhchandani, Hershleifer and Welch’s

information-based herding model. he suggested each agent has the choice to acquire

information and to choose how precise he wants his information .the author showed

that only the first agent in the given order has the incentives to acquire information

when his actions is followed by all later agents. Herd is said to have started from the

second agent.

Chen (2007) examines institutional herding behavior and its impact on the stock

market. He documented that herd by institutions usually lasts for more than one

quarter and that herd occurs more frequently for small and medium size stocks. The

author suggested that institution herding may destabilize stock prices.

Safi (2013) studied to evaluate herd behavior. He used stock price data on a daily

basis from 18 sectors for Pakistan’s market. The author uses two models that are

based on cross-sectional dispersions of stock returns. He came across with the idea

that there is no indication of herd information, market wide or industry wide at the

time of great market fluctuation by using the methodology of Christie and Huang

(1995). There was no indication of herd behavior in the sectors even by using the

methodology of Chang et al (2000). When the market was separated into upward and

11
downward market movement, overall the results show that herd behavior does not

exist in Pakistan’s Stock Market.

Wang and Canela (2007) study the herding behavior in emerging market and in

developed market. They came to know that higher level of herding is found in

emerging market than in developed markets. Finally he said that higher correlation of

herding between two markets from the same group than those from different groups.

Quiamzade and Huillier (2009) suggests that when people do not have information

about other investors they make inferences abut other investors it can be biased and

when people believe that they must be done by someone who has better information

about it. They find that people herd when they have unambiguous information.

Herding is also followed by analysts especially when they revise their

recommendations.

Market reaction is strong towards the analyst’s recommendation revision when the

recommendations are different from the consensusand reaction will be weak when

they move towards the agreement for that stock (Jegadeesh and Kim, 2008).

Previous researches have shown that people are influenced by others when making

decision

Natividad, Pilar and Sandra, (2009) by using the methodology offered by Christie and

Huang (1995), examine intentional herding and compared the dispersion in particular

market with the standard deviation of unnaturally produced market and concluded that

Spanish market show signs of inclination towards herding.

12
Khan (2011), explains herding on the basis of several factors i.e. market performance,

size etc. this empirical study showed that herding behavior is present in all countries

excluding the periods of market disorder and disaster.

There is a growing trend of herding behavior in financial markets on the determining

its origin, crashes and measuring the stability of markets. Herding behavior in

financial markets analysis practical and theoretical literature is produced in different

direction. Most financial market participants actually between theoretical research and

focus their attention on the causes of herding behavior. Irrational (Devenow & Welch,

(1996)) and logical: two concepts to explain the origins of herding behavior were

developed.

As it was before the new finance paradigm, "behavioral finance" is no longer in

dispute. The theories of human behavior and market prices of shares that can drive the

decision-making process have come to recognize its effects.

Rational and irrational herding behavior of investors behind the support is not very

theoretical. Investors keep following others until the information is lost (Banerjee,

1992;Welch, 1992 Bikhchandani, Hirshleifer).

Another important factor in favor of herding is Fashion (Barberis and Shleifer, 2003).

There are diverse indications of herd behavior in the market. Stocks of pension funds

have no indication of herd behavior (Lakonishok, Shleifer, and Vishny, 1992).

Herding may also be present in pension fund stocks. Evidences support that herd

behavior is present is Chinese market up to some extent (Wermers, 1999), many

aspects of asset pricing in the Chinese market parts (including, for example, from

13
different angles have been examined (Poon et al., 1998;Fernald & Roger,2002; Sun

and Tong, 2000).

The above research produces mixed result. Some of the studies shows that herd

behavior exist in stock market. While many concluded that herd behavior does not

exist. Even changes in measurement model also change the result. There is very little

study on herding behavior in Pakistan, stock market.

2.7. Hypothesis

After reviewing the previous study the following hypothesis is derived;

H: Herding behavior does not exist in Pakistani stock market.

2.8. Significance of understanding herd behavior

The above discussion and review of the previous studies on herd behavior in financial

surroundings already explains that why it is important to understand the concept.

These reasons are undoubtedly explained below;

 It gives us understanding that how herd behavior influence the asset prices.

 The study on herding behavior enables us to recognize influences on the

investor’s decision making process. (Sardjoe, 2012)

Human psychology is required to give details that why investors take no notice of

their past ideas and attitudes and go behind other people blindly (Devenow & Welch,

1996).

Moreover ,herd behavior is a phenomena which can be occur at intraday level when

some new information came to market ,investors and managers have not enough time

14
to apply complex models to forecast the price fluctuation for the future and therefore

their decisions are quite different from that of the investors who take their decisions

rationally after checking the models (Orlean, 1995), they imitate the other investors

without wasting their time, especially when there is extreme market situation (Henker,

Henker and Mitsios, 2006)

It is the world of ambiguity, rationality and wisdom is bounded as we are running

with informational restrictions. So because of this bounded rationality theorems and

algorithms could not completely guide in the decision making.

Herding is a fast decision making process through which people follows the actions of

others

Because of the rational behavior that other know more about long term and short term

investments and the values of assets.

Grinblatt Titman and Wermer, (1995) study herd behavior and take a sample of 274

mutual funds in the time period of 10 years and find that there is a little indication of

herd behavior. Most of them were active investors but they do not follow each other’s

actions. Another study examine pension funds and many other institutions of US

market which includes insurance companies, mutual funds, commercial banks and

mutual funds and declare that the managers of pension funds are often busy in

feedback trading slightly than others (Badrinath & Wahal, 2002).

Jones, Lee and Weis, (1999) observe that managers of pension funds take their actions

like a feedback traders particularly when they buy small stocks. They used similar

sample as Badrinath and Wahal (2002) used.

15
 CHAPTER # 03

3. DATA AND METHODOLOGY

3.1. Data

We use daily stock price data to examine existence of herding behavior in Pakistan’s

stock market. The data is collected from all the major sectors of the economy. To be

specific we include firms from 18 sectors of the economy. The sample covers a period

from 2001 to 2009. We use firms that are part of the Karachi stock exchange 100

index (KSE-100). These are the most liquid stocks and are the most traded stocks of

the market. The daily stock price data for these firms are sourced from the online data

base of business record, a premier business newspaper of Pakistan. This website

maintains a database of the daily closing prices of all the stocks traded on the Karachi

Stock Exchange. Our analysis includes the daily closing prices of the stock. We also

use B-month T-Bill rates to calculate excess stock returns.

Various researches on herd behavior in stock markets carried out herding tests on

sectors. We position 154 stocks to eighteen sectors which comprise Chemical,

Engineering Companies, Paper & Board Products, Glass &Ceramics, Technology &

Communication, Woolen, Pharmaceuticals, Power Generation, Refinery, Sugar,

Synthetic & Rayon, Vanaspati, Cement, Commercial Banks, Oil & Gas Exploration,

Oil & Gas Marketing, Fertilizer And Auto.

3.2. Measurement of Herding

For observing herd behavior a study proposed a method via cross-sectional data on

stock returns. This chapter uses model proposed by Christie and Huang (1995), this

16
model suggest that market participants use the process of investment decision-making

depend on the whole market situation. According to the asset pricing model,

individual investors invests on the basis of their own personal information and

dispersion in cross sectional returns increases with the increase in Absolute value of

market return. But, at the time of upward and downward movement of market, some

investors hold back their own personal information; mostly they mimic the communal

actions. So, it was observed that herd behavior is more prominent during bullish and

bearish trend. This model is called cross-sectional standard deviation. Christie and

Huang (1995) define the cross-sectional dispersion as;

n
CSSD t =∑ (r i , t−r p ,t )² ………………. (1)
i=1

Where

CSSD t =¿Cross -sectional dispersion of returns, for day t.

r i , t = daily stock returns for company on day t and is measured using equation (3).

r p , t = daily returns for an equally weighted portfolio return for a sector and is

calculated using equation (4)

n= number of stocks in a portfolio.

Basis behind this calculation is that, individual asset return will move in line with the

overall market returns in the occurrence of herding behavior because investor is

following the other investor’s behavior and act in accordance with the market. Due to

this reason, cross sectional standard between the stocks would be lesser than the

normal dispersion. This will indicate the existence of herding behavior.

17
Traditional asset pricing model differ in their compassion to market movements that a

cross-sectional dispersion will be increase with absolute aggregate market returns

because assets are change in their in their compassion to the market fluctuation.

Therefore, Measures for the dissemination of market closeness of asset returns mean

the return of all stocks move similar to market returns; there will be no cross-sectional

dispersion. Average market returns of individual stocks, the dispersion increased

levels vary.

According to Christie and Huang model that during extreme market condition mostly

investors restrain their private information in support of market consensus. So, herd

behavior would become more. The models for asset pricing have different predictions

and different calculations for cross sectional variation in assets [Link] pricing

models evaluate that cross sectional variation would be higher during upward and

downward movement of the market. It is because asset’s sensitivity is different to

market cumulative changes. Contrary to this herd behavior recommend that variation

will be lower when upward and downward movement of market is high and the

chances of investors to follow other investors would be high. For this reason, we have

the following equation by C & H models of large market movements during the

market test for the presence of herding.

L L U U
CSSD t =∝+ γ Dt +γ Dt + ε t (2)

L
Dt = Dummy variable and is equal to 1 when daily market returns (calculated using

equation (5)) falls in extreme lower limit (5 percentile).

U
Dt =¿ Dummy variable and takes a value of 1 when market returns (calculated using

equation (5)) falls in the extreme upper (5 percentiles) of the return distribution.

18
These two dummy variables measures the average returns for two extreme market

conditions and the average returns for the rest of the period and then the difference

(mean difference) between these two extreme market conditions and the rest of the

market and reports statistical [Link] t is measured by equation.

3.3. Sample

There are a total of 154 companies from 18 sectors of the economy. The list of the

sectors and the number and names of the companies are provided in Appendix A.

3.4. Calculation of daily stock returns

We follow standard procedure to calculate daily stock returns for each firm using the

following formula:

Returnk =ln
( )
Pt
Pt −1
−−−(3)

Where:

Return kIs daily stock return for firm k at time t.

Pt : It is daily closing price for firm k for day t

Pt −1 : It is daily closing price for firm k for previous day relative to day t

ln : Ln stands for natural logarithm.

3.5. Construction of equally-weighted portfolio of stocks

To examine the presence of herd behavior across market, I construct equally-weighted

portfolio of stocks for each sector. In an equally-weighted (or un-weighted) portfolio,

19
all stocks carry equal weight regardless of their price and market capitalization. The

actual movements in the value and returns of the portfolio are based on the arithmetic

average of the percentage changes in the price of the stock which is part of the

portfolio (Reilly and Brown, 2011)1. Typically, an equally-weighted portfolio return

can be calculated using the following formula:

Portfolioreturnt = ( % pricechange +% pricechagne

1
N
+ … … .+% pricechange
2
) … (4) n

N is the total number of firms in the portfolio.

I construct equally-weighted portfolio of stocks for all the 18 sectors in the sample.

This equally-weighted portfolio is then used to construct herd behavior measures for

empirical examination of the presence of herd behavior in Pakistan’s market.

3.6. Calculation of market returns

I use the daily closing values of the Karachi Stock Exchange 100 Index (KSE-100

Index) as a proxy for market returns. There are three stock indices in Pakistan’s

market. These are KSE-100 index, KSE All Share Index and KSE-30 Index. KSE-All

Share Index comprises all stocks listed on the Karachi Stock Exchange. Since almost

half of the shares are not actively traded on the stock exchange as some of the

companies are either non-operative or have been delisted or their stocks are not traded

even for several months, KSE-All share index is not a representative of the market

and is, therefore, not widely followedby market participants. On the other hand, KSE-

30 index comprises only 30 stocks and, hence, a very narrow index of the market.
1
Frank K. Reilly and Keith C. Brown (2011).Investment Analysis and Portfolio Management ISBN-
13: 978-0538482387 ISBN-10: 0538482389 Edition: 10th

20
KSE-100 index comprises 100 largest market capitalization stocks and is the most

followed index of the market. It is a value-weighted index and represents more than

80 per cent of the market capitalization. Moreover, previous studies on Pakistan’s

market have used KSE-100 index in their analysis. The daily returns for the KSE-100

Index are calculated using the same procedure as outlined in equation (1).

MarketReturnst =ln
( )
KSE−100 Index t
KSE −100 Index t−1
−−−−−−−−−−−−−−−(5)

Where:

MarketReturnst : It is the daily returns for KSE-Index

ln : Ln stands for natural logarithm

KSE−100 Index t : It is the closing value of KSE-100 Index for day t

KSE−100 Index t −1: It is the closing value of KSE-100 Index for previous day relative

to day t

21
 CHAPTER # 04

4. EMPIRICAL RESULTS

This chapter provides analysis of returns for all sectors, cross sectional return

dispersion and the results of the regression mode that was that was discussed in the

previous chapter.

In Table: 1 descriptive statistics for mean daily log return, median, standard deviation

of returns for different sectors are given. This table illustrate that the mean return for

all sectors are positive but 5 sectors commercial banks, synthetic and rayon, power

generation and distribution, technology & communication have daily mean return

volatility.

Besides mean table .2 also shows that daily stock returns for some sectors are

positively skewed while they are negatively skewed for some sectors. Highly mean

daily returns are observed for woolen, followed by chemicals then by Oil and Gas

Exploration. While highest volatility (standard deviation) is also observed in woolen

sectors. Jerque- Bera statistics tests for normality of returns distribution. As shown by

the statistically significant values of the Jerque- Bera test for all the sectors that the

daily stock returns distribution is not normally distributed, a phenomenon normally

associated with financial time series data.

Table. 2 reports univariate summary statistics for measure of daily cross sectional

return dispersion (CSSD) for all sectors. We find maximum cross sectional return

dispersion for power generation (9.5034) followed by Pharmaceuticals (0.0212) and

the lowest dispersion is of Oil & gas marketing sector, which is about 0.0109. When

22
we compare the maximum and minimum values of daily CSSD for each sector it

shows that woolen sector has the highest (25.2565) value, at the same time fertilizer

sector has the lowest maximum (0.0840) value. Many sectors have zero minimum

value that shows there was no trading in those days in the stocks of that sector. It is

clear from the table that CSSD for all sectors show evidence of significant positive

kurtosis and skewness and Jerque bera shows significant departure from normality.

Table 1: Descriptive Statistics for mean daily log-returns of equally-weighted portfolio

of stocks for different sectors

TAUT Engineer Commerci Synth. &

Cement Chemicals
O ing al Banks Rayon
0.00010
Mean 0.000803 0.000136 -0.000189 0.000486 -0.000487
4
0.00028
Median 0.000804 -0.000437 0.001292 0.000179 -0.000613
6
0.06562
Maximum 0.069783 0.101358 0.077377 0.161381 0.220525
9
-
-
Minimum 0.05951 -0.144754 -0.091678 -0.072183 -0.153152
0.067257
1
0.01582
Std. Dev. 0.014726 0.023563 0.01906 0.015449 0.031384
5
-
-
Skewness 0.14477 -0.002271 -0.565859 0.914553 0.516693
0.065616
3
4.57193
Kurtosis 4.764655 5.133487 5.32701 13.45751 7.420701
6

Jarque- 137.747
168.8254 245.4173 269.5043 6076.683 1111.248
Bera 4
Probabilit
0 0 0 0 0 0
y

Observati
1294 1294 1294 966 1294 1294
ons

Power
Glass& Generatio
Paper & Pharmace Sugar and
Woolen Ceramic n&
Board uticals Allied
s Distributio
n

23
 0.11474
Mean 8.54E-05 0.000114 0.000532 -0.0000470 0.000439
7
Median 0 3.00E-05 0.000308 0.000266 -0.000475 3.27E-06
5.86105
Maximum 0.098184 0.079532 0.061759 0.117962 0.090938
7
- -
Minimum -0.072934 -0.050142 -0.091434 -0.063082
17.8590 0.116613
1.71796
Std. Dev. 0.018822 0.015324 0.010949 0.021457 0.016492
2
- -
Skewness 0.207779 -0.128947 0.35522 0.601432
1.38597 0.086887
19.1964
Kurtosis 6.836316 6.904114 5.505122 7.178882 6.816795
5

Jarque- 8212.76
795.1371 831.1142 341.9474 968.7629 863.4648
Bera 4
Probabilit
0 0 0 0 0 0
y

Observati
730 1294 1294 1294 1294 1294
ons

Technolog
Oil & Gas
Refiner Fertilize Oil & Gas y& Vanaspati
Exploratio
y r Marketing Communi and Allied
n
cation
0.00001
Mean 0.000327 0.000478 0.000172 -0.00013 0.000642
59
0.00014
Median 0.000753 0.000921 0.000855 0.000438 0
9
0.08667
Maximum 0.078079 0.095267 0.086376 0.103248 0.134958
1
-
-
Minimum 0.08228 -0.09951 -0.065832 -0.229615 -0.066793
0.074816
3
0.02182
Std. Dev. 0.017047 0.021142 0.017596 0.021803 0.019047
5
-
-
Skewness 0.03396 -0.292992 -0.274562 -1.158146 0.483626
0.148974
7
3.80978
Kurtosis 4.977641 4.512801 4.583019 14.83745 6.134194
1

Jarque- 35.6044
195.4921 106.2647 151.3703 7844.354 580.0759
Bera 5
Probabilit
0 0 0 0 0 0
y

Observati
1294 1173 969 1294 1294 1294
ons

24
Table2: Descriptive Statistics for CSSD for different sectors time period:

Medi Min Std. Ske Kurt J- pro

Sector Mean an Max. i. Dev. w. osis N Bera b.
Oil and
Gas
Marketin 0.009 0.086 0.00 0.007 2.39 12 15406 0.0
g 0.0109 4 8 00 2 41 19.21 94 .35 00
Automob 0.020 0.179 0.00 0.011 5.32 12 18528 0.0
iles 0.0216 5 4 11 5 83 60.65 94 7.6 00
0.020 0.325 0.00 0.014 8.37 156.2 12 12819 0.0
Cement 0.0230 1 9 25 4 58 8 94 68 00
Commer
cial 0.020 0.297 0.00 0.016 6.93 96 29041 0.0
Banks 0.0230 3 1 03 8 26 86.80 6 0.6 00
Chemical 0.028 0.624 0.00 0.022 15.3 401.6 12 86209 0.0
s 0.0316 3 1 00 1 820 8 94 11 00
Engineer 0.029 0.227 0.00 0.020 2.57 12 9648. 0.0
ing 0.0340 3 6 00 0 06 15.35 94 111 00
0.012 0.084 0.00 0.009 1.88 11 3399. 0.0
Fertilizer 0.0142 0 0 00 4 40 10.46 69 567 00
Glass
and 0.028 0.320 0.00 0.025 3.97 12 48038 0.0
Ceramics 0.0330 0 8 00 1 05 31.77 94 .54 00
Oil &
Gas
Explorati 0.012 0.196 0.00 0.011 5.04 96 18598 0.0
on 0.0139 0 0 00 5 28 70.12 9 2 00
Paper
and 0.019 0.195 0.00 0.024 3.22 12 11848 0.0
Board 0.0255 5 9 00 2 93 16.34 94 .58 00
Pharmac 0.020 0.135 0.00 0.011 2.85 12 20060 0.0
euticals 0.0212 2 8 00 0 42 21.42 94 .23 00

25
Power
Generati 3.295 0.710 0.00 1.085 35.9 1291. 12 89774 0.0
on 9.5034 2 0 00 7 174 37 94 392 00
0.015 0.123 0.00 0.012 2.35 12 8879. 0.0
Refinery 0.0183 9 9 00 6 12 14.94 94 479 00
Sugar
and 0.035 0.133 0.00 0.017 1.47 12 1127. 0.0
Allied 0.0387 2 4 35 0 65 6.49 94 144 00
Synthetic
and 0.036 0.321 0.00 0.036 1.97 12 3506. 0.0
Rayon 0.0451 6 6 00 9 79 10.03 94 604 00
Technolo 0.022 0.505 0.00 0.019 12.7 282.4 12 42447 0.0
gy 0.0250 7 2 00 8 337 3 94 86 00
Vanaspat
i and 0.024 0.283 0.00 0.021 2.48 12 18666 0.0
Allied 0.0274 5 6 00 8 83 20.93 94 .84 00
1.003 25.25 0.00 0.974 3.06 73 24108 0.0
Woolen 1.6131 0 65 00 9 83 30.48 0 .99 00

4.1. Regression model results for herd behavior

We use the model of return dispersion proposed by Christie and Huang (1995) to

examine herding. We calculate the results of this model reported in Table 3. We use

lower and upper 5 percentiles of index return as a period of market stress. The given

table shows that there are some sectors which have positive value of coefficients ( γ L

andγ U ) but some sectors have negative values. The significant and positive values

show that during the period of extreme market movement, there is no herd behavior in

those particular sectors. The positive and significant dummy coefficients ( γ L andγ U )

in this table indicate that the return dispersion of stocks increase during the time

period of large market movement. The negative values of dummy coefficients indicate

that there is lower volatility or dispersion than the market i.e. the price movement is

not highly correlated with the market. Some of the sectors show little evidence of

herd behavior i.e. oil and gas exploration, paper and board, refinery, sugar and allied,

technology and communication, vanaspati and allied industries show signs of herd

behavior during downward market movement and in woolen sector it is found in both

26
upward and downward market movement. On the whole, we can say that there is no

indication of herd behavior. R-squared is the coefficient of determination; it is the

percentage of the response variable variation that is explained by the linear model.

Table 3

Regression Analysis for CSSD

Industry α γ
U
γ
L Adj.R2 F-test

Cement 0.021473 0.02747 0.192318 0.03474 236.8785

 Standard Error 0.000608 0.004137 0.006723

 T-stat 35.31743 6.640077 28.60598

 Prob. 0 0 0

Fertilizer 0.013772 0.001288 0.021684 0.05748 21.6279

 Standard Error 0.000488 0.003762 0.004118

 T-stat 28.22131 0.342371 5.265663

 Prob. 0 0.7755 0

27
Oil and Gas Exploration 0.013321 0.1219 -0.00934 0.04728 10.3509

 Standard Error 0.000521 0.007298 0.004636

 T-stat 25.56814 16.70321 -2.01467

 Prob. 0 0 0.035

Oil and Gas Marketing 0.014077 0.001728 0.030068 0.0528 21.7035

 Standard Error 0.000473 0.004691 0.006619

 T-stat 29.7611 0.368365 4.54268

 Prob. 0 0.7589 0.0002

Commercial Banks 0.02309 0.000618 0.107717 0.0648 27.1635

 Standard Error 0.000803 0.008517 0.005609

 T-stat 28.75467 0.072561 19.20431

 Prob. 0 0.9518 0

Automobiles Assembler 0.020093 0.016114 0.045482 0.06 21.6825

 Standard Error 0.000629 0.004339 0.003211

 T-stat 31.94436 3.713759 14.16443

 Prob. 0 0.002 0

Engineering 0.033689 0.008735 0.000748 0.0336 13.5345

28
  Standard Error 0.001154 0.00354 0.001194

 T-stat 29.19324 2.467514 0.626466

 Prob. 0 0.04 0.602

Glass and Ceramics 0.025961 0.006511 0.002589 0.096 53.907

 Standard Error 0.001166 0.004371 0.002244

 T-stat 22.26501 1.465314 1.552348

 Prob. 0 0.1431 0.1208

Paper and Board 0.025825 0.003228 -0.00344 0.0456 18.7215

 Standard Error 0.00108 0.004269 0.001441

 T-stat 23.91204 0.756149 -2.38723

 Prob. 0 0.5288 0.0468

Pharmaceuticals 0.020704 0.000384 0.001673 0.0348 14.5845

 Standard Error 0.000692 0.001955 0.000659

 T-stat 29.91908 0.196419 2.538695

 Prob. 0 0.87 0.0347

Refinery 0.017281 0.005077 -0.00043 0.0576 23.9715

 Standard Error 0.000651 0.002225 0.00075

29
  T-stat 26.54531 2.281798 -0.57333

 Prob. 0 0.0415 0.6363

Sugar and Allied 0.033556 0.001309 -0.00092 0.0936 39.081

 Standard Error 0.001154 0.002949 0.000997

 T-stat 29.07799 0.443879 -0.92277

 Prob. 0 0.7114 0.4391

Synthetic and Rayon 0.042097 0.015944 0.002903 0.0504 20.958

 Standard Error 0.001708 0.006494 0.002194

 T-stat 24.64696 2.455189 1.323154

 Prob. 0 0.041 0.2705

Technology and Communication 0.025915 0.004726 -0.00019 0.0336 8.9145

 Standard Error 0.000951 0.003528 0.00119

 T-stat 27.25026 1.339569 -0.15966

 Prob. 0 0.2646 0.8948

Vanaspati and Allied Industries 0.028771 0.000971 -0.00072 0.0324 7.9695

 Standard Error 0.001033 0.003896 0.001318

 T-stat 27.85189 0.24923 -0.54628

30
  Prob. 0 0.8355 0.651

Woolen 1.328026 -0.46855 -0.07326 0.0384 7.2345

 Standard Error 0.94789 0.987001 0.136784

 T-stat 1.401034 -0.47472 -0.53559

 Prob. 0.346 0.392 0.156

Chemicals 0.033584 0.011438 0.001152 0.0228 5.8695

 Standard Error 0.001134 0.003955 0.001334

 T-stat 29.61552 2.892035 0.863568

 Prob. 0 0.0161 0.4721

Power Generation and Distribution 0.046648 0.018715 0.21926 0.0012 1.869

 Standard Error 0.067265 0.356529 0.120232

 T-stat 0.693496 0.052492 1.823641

 Prob. 0.5634 0.9651 0.1288

Appendix A: List of companies

Glass And Paper And Technology And

Chemicals Engineering Woolen
Ceramics Board Commun.

1 BIAFO ADOS Baluchistan Baluchstan Callmat Bannu

31
 Woolen

2 BAWANY Bolan Emco Century Humnet Moonlight

3 BERGER Crescent Frontier Cherat Netsol

4 BOC DADEX Ghani Merit Pakdata

5 BUXLY Dost Karam Packages Ptcl

6 COLGATE Gauhar Medi Pkpaper Telecard

7 DYNEA Hufaz Sabri Security Trg

8 ICI Intel Tariq Worldcal

9 ITEHAD KSB Callmat

10 NIMIR Metropoliton Humnet

Pak
11 DESCON Netsol
Engineering

12 PAKGUM PAKDATA

13 PTA PTCL

14 SARDAR TELECARD

32
 SITRACHE
15 TRG
M

16 WAH WORLDCAL

Pharmaceu Power Synthetic And

Refinery Sugar Vanaspati
iticals Generation Rayon

1 Abbot Kapco Attock Adam Pksyn Kaka khel

2 Ferozsons Kohinoor Byco Abbas Polyron Punjab

3 Glaxo Kesc National Abdulah Ropaly Wazir

4 Highnoon Sitara Pak Noor Tristar Sss

5 Otsuka Tristar Ansari

6 Sanofi Altern Bawany

7 Searle Genertec Chashma

8 Wyeth Japan Crescent

Kohinoor
9 Dewan
Power

10 Sgpowr Faran

11 Sthelctr Haseeb

33
 12 Habib

13 HabibSG

14 Jdwan

15 Mirpur

16 Mirza

17 Noon

18 Pangrio

Appendix A-Continued

Commercial Oil And Gas Oil And Gas

Cement Fertilizer Auto
Banks Exploration Marketing

1 DGKC BAFL OGDC SSGC Dawood Hino

34
 Hercules

FFC Bin Atlas

2 Lucky Akbl POL SNGP
Qasim Honda

3 Mlcf Fabl MARI APL ENGRO Dewan

Honda
4 Poic Bop PPL PSO FFC
Atlas

5 Fecto Soneri SHELL Indus

Kohat
6 MCB Millat
Cement

Cherat Paksuzuk
7 Bahl
Cement i

8 Attock NBP Alghazi

Fauji
9 Samba Gnissan
Cement

1
Dadbhoy Habib METRO
0

1
ZEALPAK NIB
1

35
1
SILK
2

CHAPTER # 05

5. SUMMARY AND CONCLUSION

36
I would like to discuss the facets that are covered in this research before answering

the problem statement. Therefore, a summary of the research is provided. In the end I

have discussed the general conclusion and limitations.

5.1. Brief summary

This research argues the herd behavior in stock market of Pakistan. For this purpose

Karachi stock exchange (KSE-100 Index) was selected which is the representative of

stock market. It was investigated that whether extreme market condition affects the

behavior of investors. Prior to examine the imitating behavior of investors, literature

was discussed in which, different kinds of herding, the measurement of herding and

the significance of understanding imitating behavior stock market were explained.

After this chapter, the details of methodology was given in which Christie and Huang,

(1995) was discussed that how it could be approached. Accordingly, regression model

was applied and results were interpreted, which go ahead to the general conclusion.

5.2. General conclusion

The unavailability of correct information and unclear investment surroundings in

emerging market induce investors to keep on herding (Chen, Rui & Xu, 2003).

Improved and understandable information would probably reduce the chance of

herding (Andersson, 2009).

In this research, herding behavior of investors in Pakistan’s market is observed by

using the sample of 18 sectors of KSE-100 index.

In a few sectors, a slight evidence of herd behavior was found during large market

movement but from overall results, it become clear that herd behavior does not exist

37
in Pakistani stock market from the time period of January 2001- December 2009 by

applying the methodology of Christie and Huang, (1995).

38