Sip+Report +Metrics4+Analytics

8/8/2019 Sip+Report +Metrics4+Analytics

1/74

Event Based Financial Research 2008

1

A REPORT

ON

EVENT BASED FINANCIAL RESEARCH UsingARMA, Single Index and Risk Adjusted model

Submitted by,Amit Kr. Jaiswal

07BS0427

Metrics4 Analytics


2/74


2

A Report

On

EVENT BASED FINANCIAL RESEARCH UsingARMA, Single Index and Risk Adjusted model

By

Amit Kr. Jaiswal

A Report Submitted in Partial Fulfillment of the Requirements of MBA

Program

Distribution List:

Mr. Anjaneyulu Marempudi(Founder & CEO, Metrics4 Analytics)Mr. Rajeev Gupta (Director - Research & Analytics, Metrics4 Analytics)

Mr. Sanjay Banka (Director - Research, Metrics4 Analytics)

Dr. Jagrook Dawra (Professor, ICFAI Business School, Hyderabad)


3/74


3

Acknowledgement

I express my sincere gratitude to CEO of Metrics4 Analytics, Mr. Anju Marempudi for his

encouragement, support and valuable guidance throughout the project duration. In spite of being

fraught with unending engagements in office, he kept me motivating to try best at all times.

I am also thankful to my Company Guide, Mr. Rajiv Gupta for not only providing me valuable

guidance and support for project but also providing me the required support with his extra

ordinary knowledge of IT.

The project area was entirely unknown to me. It required a lot of knowledge and guidance. I

would also like to express my gratitude to the Analysts Team at Metrics4 Analytics for

constantly elucidated upon my repetitive queries.

I would also like to thank my faculty guide, Prof. Jagrook Dawra for providing me with his

constant support and guidance in preparing this report.

I would also like to thank Renjith Sivaram, student of IBS Hyderabad, Batch of 2008, for

providing me support in application of ARMA, in spite of his busy work schedule.

Prof. Chakrapani deserves a special mention as he has consistently enlightened me with his

knowledge and experience of Event Analysis, which was a great help to me.

Lastly, I would like to thank ICFAI Business School and Metrics4 Analytics for providing me an

opportunity to gain hands-on experience by working in a corporate environment.


4/74


4

Table of Contents

1. Abstract8

2. Project Title...10

3. Objectives of the Project...10

4. Introduction..10

4.1What is an Event?.......................................................................................................10

4.2What is the importance of an Event?..........................................................................11

4.3What is Event Analysis...12

4.4Analysis of Major Events4.4.1 Earnings Release.12

4.4.2 Dividend.14

4.4.3 Guidance.16

4.4.4 Legal/ Regulatory...17

4.4.5 Product Launch..18

4.4.6 FDA Filings21

5. Review of Literature .23

5.1An Empirical Analysis of Reactions to Dividend Policy Changes for NASAQ

Firms...24

5.2An Empirical Study On Stock Price Responses To The Release Of The

Environmental Management Ranking In Japan..25

5.3Announcement Effect And Price Pressure: An Empirical Study Of Cross Border

Acquisitions By Indian Firms 26

5.4Are retailing mergers anticompetitive? An event study analysis................................27

5.5Vindication Of The Event Driven Investment Strategy..28

5.6Corporate Restructuring In Japan: An Event Study Analysis.29

6. Price Prediction Using Box- Jenkins Methodology

6.1Scope of The Analysis30

6.2Data Collection...31


5/74


5

6.3Sample31

6.4Outline of The Analysis..31

6.5Definition of Variables...31

6.6Methodology..33

6.7Application of ARMA modeling Process..36

7. Comparative Analysis Using Single index and risk adjusted Model43

7.1 Scope of The Analysis43

7.2Data Collection....43

7.3Outline of The Analysis...44

7.4Definition of Variables....45

7.5Methodology45

7.6When companies Make Dividend surprise..47

7.6.1 Key Findings and Conclusion47

7.7When companies Make Dividend surprise (0 to 10% increase).51

7.7.1 Key Findings and Conclusion.51

7.8When companies Make Dividend surprise (10 to 20% increase)56

7.8.1Key Findings and Conclusion.56

7.9When companies Make Dividend surprise (20 to 50% increase)62

7.9.1 Key Findings and Conclusion.62

8. Cross Sectional regression.67

8.1Scope of The Analysis.67

8.2Data Collection....67

8.3Outline of The Analysis...68

8.4Definition of Variables....68

8.5Methodology68

8.6Result of regression.69

9. Limitation of Study....73

10.Scope of Further Study..74

11.References..75


6/74


6

Table of Illustrations

1. Figure 1: Earnings Release of Lehman Brothers..13

2. Figure 2: Dividend Declaration by Ambac....14

3. Figure 3: Guidance Issued by Countrywide Financial..16

4. Figure 4: Class Action Lawsuit Faced by Bear Sterns......18

5. Figure 5: Waters AquaAnalysis System Launched by Waters Corporation.....19

6. Figure 6: FDA Approval Received by Mentor Corp21

7. Figure 7: In- Sample prediction of stock Price for the period of 1 month39

8. Figure 8: Abnormal return during event window (Single Index Model)...47

9. Figure 9: Abnormal return during event window (Risk Adjusted Model) ...47

10.Figure 10: Cum. Abnormal return during event window (Single Index

Model)48

11.Figure 10: Cum. Abnormal return during event window (Risk Adjusted Model).48

12.Figure 11: Abnormal return during event window ...35

13.Figure 12: Abnormal return during event window .......45

14.Figure 13: Abnormal return during event window ...4515.Figure 10: Cumulative Abnormal return during event window 46

16.Figure 11: Cumulative Abnormal return during event window 46

17.Figure 12: Abnormal return during event window.53

18.Figure 13: Abnormal return during event window.53

19.Figure 14: Cumulative Abnormal Return54

20.Figure 15: Cumulative Abnormal return during event window 54

21.Figure 16: Abnormal return during event window (10 to 20%) 58

22.Figure 17: Abnormal return during event window (10 to 20%)..58

23.Figure 18: Cumulative Abnormal return during event window..59

24.Figure 19: Cumulative Abnormal return during event window.59

25.Table 1: Correlogram of Stock prices of IMCL..36

26.Table 2: Test of Unit Root at 1stdifference...37


7/74


7

27.Table 3: Correlogram of IMCL Stock price after application of ARMA Modeling

process...........................................................................................................................................38

28.Table 4 :Table values after application of Estimation Model..........39

29.Table 5: Model Accuracy check for in-sample prediction41

30.Table 6 : Calculation of Significance of Mean Abnormal Return (Single Index

Model)........50

31.Table 7 : Calculation of Significance of Mean Abnormal Return (Risk adjusted

Model)........51

32.Table 8: Calculation of Significance of Abnormal return (Single Index

Model)........................55

33.Table 9: Calculation of Significance of Abnormal return (Risk adjustedModel)..56

34.Table 10 : Calculation of Significance of Mean Abnormal Return .......60

35.Table 11 :Calculation of Significance of Mean Abnormal Return ......61

36.Table 12 :Calculation of Significance of Mean Abnormal Return ......65

37.Table 12 : Calculation of Significance of Mean Abnormal Return66


8/74


9/74


9

help them accordingly to take his/ her position. Till the completion of this report I could develop

the model for Insample prediction with 99% accuracy but out sample prediction still require

some more refinement of model and understanding of topic.

To identify the financial factors which affect the stock return a Cross sectional regression

analysis has been undertaken by taking Cumulative abnormal return of 1 day event window as

dependent variable and Percentage Increase in dividend, dividend payout ratio, dividend yield,

P/E ratio as dependent variable. Result suggest that even though these factors significantly affect

the return but there intensity of impact is very small and leads to conclusion that these factors

dont form part of the investment decision of event driven investor.


10/74


10

PROJECT TITLE

Event Based Financial Research using ARMA, Single Index and Risk adjusted model

OBJECTIVES OF THE PROJECT

To establish and substantiate the impact of major events on stock prices.

To find out the time line of effect of major event types.

To Compare and substantiate the result through the application of different Models.

INTRODUCTION

WHAT IS AN EVENT?

An event is a corporate action initiated by a public company that affects the securities (equity or

debt) issued by the company. Any announcement made by the company, regarding its businessand operations may become an event. Also, anything said or reported for a company by any

person, magazine or journal may become an event depending upon the reliability and popularity

of the source from where it has been generated.

WHAT IS THE IMPORTANCE OF AN EVENT?

An event acts as a raw material for an investor on the basis of which he makes his final product,

i.e., his decision of what he has to do next with the stock of the related company. An event isimportant information, or an opportunity which can be cashed on by the investor to earn more

than normal returns from a stock.


11/74


11

WHAT IS EVENT ANALYSIS?

Event analysis is a technique to form an Event Driven Investment Strategy. An event driven

investment strategy is one which aims to capitalize on the irrationality of the investors. An event

contains certain information which may lead to rise in fall of the stock prices. So, it becomes

utmost important to analyze the event completely before an action is taken on it. This is where

event analysis comes into picture. An improper or incomplete analysis of an event may lead to an

unwise decision which may lend the investor into trouble. At the same time, a proper event

analysis can help an investor to earn higher returns than he does in the usual course of investing.

Another advantage which an event driven investment strategy has over other strategies like

technical analysis, diversification, etc., is its simplicity. Most of the techniques involve a hugeamount of probability and estimation whereas this strategy depends upon having a firm grasp on

financial instruments and their potential risks and profits. All that an investor needs to focus is

availability of right information at the right time in the right format.


12/74


12

Analysis of Major Events

To elaborate more about events, let us consider the following examples of major events which

affect a companys stock prices. There are 72 event types identified by the company ranging

from as important as earnings release to as diversified as impairment charges. However, to build

a basic understanding of events the following examples would suffice.

EVENT 1

Tue Mar 18, 2008: Lehman Brothers Holdings Inc. reported results for 1Q ended Feb 29,

2008. Net revenues dipped 31% YoY to $3.5 billion. Net income was $489 million (EPS

$0.81), down from $1.15 billion (EPS $1.96) a year ago. Current quarter net revenues

reflect negative mark to market adjustments of $1.8 billion.

INVESTORSSENTIMENT

The present US financial Sector is characterized by heavy losses by worlds biggest financialinstitution and Investment Bankers. The case became worse with the fall of worlds 5

th largest

investment banker Bear Sterns. All these instances shacked the confidence of US investors on

these financial giants and consequently there stock prices plummeted .The same case happened

with Lehman Brothers, the stock of company were shooting around its bottom because of heavy

loss expectation by the Investors, and the same confirmed by company in its Guidance.

Following this expectation company reported a decrease of 57 per cent in its net profit over the

corresponding quarter of the previous year; still the stock price went up by almost 60% beating

the S&P index. This shows that the investors were expecting Lehmans profit to go down by

more than 57% or may be the result turns out to be better compared to other companies in the

same league.


13/74


14/74


14

EVENT 2:

Wed Jan 16, 2008: Ambac will reduce its quarterly common stock dividend from $0.21 pershare to $0.07 per share.

INVESTORSSENTIMENT:

Unlike in India, in the US, the companies declare quarterly dividend. The dividend in itself does

not create any significant volatility either in stock return or in its trading volume (the same has

been proved in event research undertaken). But any dividend surprise attracts the attention of

investors and consequently generates the volatility in stock price as well as trading volume.

These dividend actions signal the investor about the companys future prospect. Here Ambac

Corp. was in loss from the past two quarters because of the heavy impairment charges and to

strength its financial position the Board of Directors decided to reduce the dividend amount but

this signaled to the market wrongly and stock crushed towards its bottom.

EVENT DRIVEN INVESTMENT STRATEGY

Figure 2: Dividend Declaration by Ambac

Source: http://www.eventvestor.com/evp_event.php?eid=139800


15/74


15

Past quarter losses and falling EPS had already cued to investors about the bad financial of the

company and resultantly stock was neutral in terms of returns in the market. But on the news of

reduction in dividend rate investors disbelief in the stock strengthen and stock fall to more low

level, and fall more on the second day of the declaration because of the spread of announcement

effect but within the days, the full information spread into the market, that company could not

deliver the result because some of the onetime charges and revenues of the company is rising

constantly, resultantly the stock shored up and outperformed the market by 52 per cent. An event

driven investor could use this situation and generate the hefty return of 120 per cent by taking the

position opposite to the market because on the day of declaration the market was very negative

and purchase on that day and selling three days after when market could get the full information,

will have substantiated his portfolio performance.


16/74


16

EVENT 3

Fri Oct 26, 2007: Countrywide Financial provided guidance for 4Q 07 & FY 08. 4Q 07EPS: $0.25 to $0.75. FY 08 Return on Equity: 10% to 15%.

INVESTORSSENTIMENT

Guidance is the forecast which a company makes about its next quarter or current fiscal or next

fiscal earnings. Guidance exhibits a companys confidence about its financial condition and in

this case in spite of the turbulent US financial sector, the company showed the positive

improvement in its earnings for the quarter as well as Full FY 08, it not only helped the company

to gain momentum by outperforming the NYSE financial Sector index by 38 per cent but also

surpassing the S&P 500 index by 30 per cent.

Figure 3: Guidance Issued by Countrywide Financial

Source:http://www.eventvestor.com/evp_event.php?eid=119725
http://www.eventvestor.com/evp_event.php?eid=119725http://www.eventvestor.com/evp_event.php?eid=119725http://www.eventvestor.com/evp_event.php?eid=119725http://www.eventvestor.com/evp_event.php?eid=119725


17/74


17

EVENT 4

Mon Mar 17, 2008: Coughlin Stoia Geller Rudman & Robbins announced a class actionsuit on behalf of purchasers of The Bear Stearns Companies Inc. stocks during the period

between Dec 14, 2006 and Mar 14, 2008. The complaint alleges that during the Class

Period, defendants made false and misleading statements regarding the business and

financials.

INVESTORSSENTIMENT

The Bear Stearns Companies Inc., the name in itself suggests the end of the era of the US

financial market. This event shows the legal issues faced by the company. In corporate world,

many companies sue or get sued by other companies for one or other reason. These lawsuits

significantly affect the companies bottom-line as they involve huge expenditures and charges.

Here the end of the Bear Sterns came all of the sudden and therefore attracted a lot of legal

action in the form of investors grievances against the company. The market was already

fumbled by the sudden fall of the company and investors loss of value as well as money

aggrieved the situation.


Figure 4: Class Action Lawsuit Faced by Bear Sterns


18/74


18


On 17th Mar the news of proposed Acquisition of Bear Sterns by JP Morgan for $4 per share

touches the Wall Street building and took the prices of Bear Sterns Share to new bottom.

Therefore this fall in prices could not attributed purely to this event, rather was a mixed reaction

of few events together. The price shoots up on 24 th Mar by making the abnormal return of 80 per

cent because of the indication by the J P Morgan to raise the offer price. In this turbulent market

the right strategy would be to see the right price and exit from the company.


19/74


19

EVENT 5

Tue Oct 23, 2007, Waters Corporation unveiled its new Waters AquaAnalysis System, atotal system solution designed to detect pesticides and other contaminants in drinking

water and meet or exceed regulatory requirements worldwide.

INVESTORS SENTIMENT:

When a company launches a new product, its earnings are expected to increase. It depends on the

novelty of the product, its utility for the customers, its direct and indirect competitors, and the

market share it is expected to command. In the case of Waters Corporation, Waters Aqua

Analysis System was much awaited product of the company. Company has been doing good in

terms of manufacturing quality products and delivery value to the customer and addition of new

high quality product in its product line, created a positive sentiment in the market and lifted the

stock price.


Figure 5: Waters AquaAnalysis System Launched by Waters Corporation


20/74


20


The launch of the product by the company acted positively for the company and lifted the stock

price on the day of launch, event this case is in line of the famous ipod case as mentioned

previously. Since the company was in process from the past few months for getting the

regulatory approval and launching the product. Since the news came to the market early, it

reduces the gain opportunity by this launch. Even though an investor acting on the sole market

news could make a return of 6 per cent above the market return, if he/she invests 1 day before

and remain invested day after the event.


21/74


21

EVENT 6

Mar 21, 2008: Mentor Corporation received FDA approval for Prevelle Silk. It is the firstof a new line of lidocaine containing hyaluronic acid dermal fillers that Mentor anticipates

to market and distribute globally.

INVESTORS SENTIMENT:

As we know that the event like FDA is more important for the companies in the healthcare

sector. But the return on these stocks depends on level of filing for approval. Here the drug adds

very important feature of the company existing portfolio, but perhaps the initial (Phase 1) level

of filing could not attract much of the interest of the investor.

Figure 6: FDA Approval Received by Mentor Corp



22/74


22

EVENT DRIVEN INVESTMENT STRATEGY:

A Phase 1 filling by the company generally does not attract much of the investors sentiment, so

long as it is a Blockbuster Drug and it is much awaited by the market. In the other cases the

investor could get in to stock before the declaration of the FDA filing by the company. Like in

this case an investor can enter 2 day before the declaration and can make a marginal return of

12% by beating the market.


23/74


23

Review of Literature

AN EMPIRICAL ANALYSIS OF REACTIONS TO DIVIDEND POLICY CHANGES FOR NASDAQ

FIRMS

Author: Patricia A. Ryan (Colarado State University), Scott Besley (University of South

Florida) and Hei Wai Lee (University of Michigan-Dearborn)

Published in: Journal of Financial and Strategic Decisions Volume 13 Number 1

Year: 2000

Objective of Study: To examine the information content of dividend policy

Insight of the Paper:

This paper circles around two arguments, the signaling argument and the free cash flow (FCF)

argument. The signaling arguments present the basis for asymmetric information between

managers and shareholders. Given this environment, management has the incentive to signal

positive firm-specific private information to shareholders. Negative information would be

withheld until financial constraints force the release of such information. FCF argument says that

managers tend to hoard cash to invest in negative NPV projects for their own utility

maximization. The agency costs that result from this overinvestment decrease the value of the

firm.

The purpose of this research is to test the dividend signaling and free cash flow hypothesis to

determine if either hypothesis better explains stock price reactions to changes in dividend policy

for NASDAQ firms.

This paper uses an event study methodology that disentangles the mean effect from the variance

effect when measuring a change in stock prices. The authors used dividend initiations and

omissions, as opposed to changes in dividend payments, in the research. This is done because


24/74


24

dividend initiations and omissions reduce the bias in estimating the dividend surprise because the

announcements are less likely to be anticipated. This allows for a stronger test of the signaling

hypothesis.

The analysis of data is done using regression analysis. Average standardized abnormal return,

and its variance is calculated. Finally, a t-test is applied to calculate the unconditional wealth

effect.

Usefulness for the Project

This research paper uses an improved event study methodology, which controls for fluctuations

in idiosyncratic risk around the announcement, to document significant wealth and variance

effects upon the initiation or omission of dividends by NASDAQ firms. This paper provides a

base for event study analysis. This paper gives a range of statistical applications which can be

used in the project.

AN EMPIRICAL STUDY ON STOCK PRICE RESPONSES TO THE RELEASE OF THE

ENVIRONMENTAL MANAGEMENT RANKING IN JAPAN

Authors: Fumiko Takeda (University of Tokyo) and Takanori Tomozawa (University of Tokyo)

Published in: Economics Bulletin, Vol. 13

Year: 2006

Objective of the Study: To analyze the impact of the release of the Nikkei EnvironmentManagement Ranking on stock prices


25/74


25

Summary

This paper takes the release of the Nikkei Environment Management Ranking as an event and

attempts to find out its impact on the stock prices of the firm. In this paper, the authors have

taken a sample of top 30 companies in this release. They have used an event study methodology

in their research. They have chosen a three-day event window and 150 day estimation window.

Then, abnormal return, cumulative abnormal return, and its variance have been calculated. Then,

the normality of CAR is checked using J-statistic.

The authors have concluded that the stock prices on the whole did not respond significantly to

the release of the ranking within a three-day event window. Moreover, stock prices of companies

that experienced a downgrade increased significantly, while those that experienced an upgrade

decreased significantly.


The authors have used event based methodology to establish a relationship between stock prices

and Nikkei Environment Management Ranking. The quantitative tools used in the research like

Cumulative Abnormal Return, J-statistic, etc. can be used in the project for better analysis and

results.

ANNOUNCEMENT EFFECT AND PRICE PRESSURE:AN EMPIRICAL STUDY OF

CROSS BORDER ACQUISITIONS BY INDIAN FIRMS

Authors: PengCheng Zhu (Carleton University) and Shavin Malhotra (Carleton University)

Published in: International Research Journal of Finance and Economics, Issue 13

Year: 2008


26/74


26

Objective of Study: To examine the stock performance of a sample of Indian firms acquiring

US firms in the period 1999-2005

Summary

Cross Border M&As have increased substantially over the last few years. The transaction value

has increased by 88% over 2004 to US $716 billion and the number of deals has increased 20%

to 6134. Although the firms from the developed world command lions share of cross border

acquisition, acquisitions from developing economies like India and China have entered the fray

in a big way. A little research work has been carried on cross border acquisitions by developing

countries, and this research paper attempts to fill this gap. The author has adopted standard event

analysis methodology and used Mean Adjusted Return Model to calculate Abnormal Return and

Cumulative Abnormal Return, to observe the market reaction to the M&A announcement event.

To avoid the possibility of effect of other factor on abnormal return, the author has used

multivariate regression with key determinant variables. The author concluded by saying that the

acquisitions of the US firms by Indian Companies have positive impact on the acquiring firms

value in the initial days after the announcement date, but the trend become negative in the next

few days in the announcement period.


This paper provides with the tools which can be used to undertake event based analysis. The

methods of use of applications like abnormal return, cumulative abnormal return, multivariate

regression analysis, etc. can be included in the project using this research paper as base.

ARE RETAILING MERGERS ANTICOMPETITIVE?AN EVENT STUDY ANALYSIS

Authors: John David Simpson (Federal Trade Commission) and Daniel Hosken (Federal Trade

Commission)


27/74


27

Published in: NA

Year: 1998

Objective of Study: To determine whether four retailing mergers that occurred during the late

1980s reduced competition

Summary: In this study, in order to examine the mergers that would have been most likely to

reduce competition, the authors drew a sample from the set of transactions that occurred between

1984 and 1993. The four mergers analyzed in this study are: May Department Stores Companys

1986 acquisition of Associated Dry Goods, Great Atlantic & Pacific Tea Companys 1986

acquisition of Waldbaum Inc., Von Companies 1987-1988 acquisition of Safeways Southern

California stores, and American Stores 1988 acquisition of Lucky Stores Inc. The authors have

used the market model to identify those dates on which the target firm experienced large and

statistically significant abnormal returns. The authors have related these mergers to the

Herfindahl-Hirschman Index (HHI) computed for the area or areas affected by the merger. The

authors have also used market model to analyze the impact of these M&As. Using this

methodology, the authors found that rival firms experienced positive abnormal returns from May

Companys 1986 acquisition of Associated Dry Goods and American Stores 1988 acquisition of

Lucky Stores. These results offered some evidence that retailing mergers that lead to large

increases in concentration in already concentrated markets may lessen competition and lead to

higher product market prices.

Usefulness for the Project: The authors in this study have also used the same market model as

used in An Empirical Analysis of Reactions to Dividend Policy Changes for NASDAQ Firms,

mentioned above. This research paper gives an insight of event study from the M&A perspective.

VINDICATION OF THE EVENT DRIVEN INVESTMENT STRATEGY

Author: Adam T. Samson


28/74


28

Published in: NA

Year: NA

Objective of Study: To explain the usefulness of an Event Driven Investment Strategy

Summary

The Event Driven Investment Strategy is one that aims to capitalize on the irrationality of

investors. This research paper includes study of earnings release Osiris Therapeutics (OSIR),

launch of Ipod by Apple Computer (AAPL), rumors regarding Arch Coal (ACI), earnings release

of Steel Dynamics (STLD), earnings release of Oakley (OO), and earnings release of Molex

(MOLX). All these companies belong to different sectors. The purpose of taking the companies

from different sectors is to generalize the impact of events on all the companies regardless of

their sectors.

The impact of these events on the stock prices of these companies is shown graphically. The

stock prices have moved up and down depending upon the investors expectations from the

companies and their performance.


This research paper forms the base for conducting an event based research. The study done in

this paper shows the impact of the events on the stock prices. There are no statistical techniques

used in this paper. The impact has been shown by graphical representation of stock price

movements only. The conclusions of this study can be proved by using other techniques as used

by other authors.

CORPORATE RESTRUCTURING IN JAPAN:AN EVENT STUDY ANALYSIS

Author: Jorge A. Chan-Lau

Published in: IMF Working Paper


29/74


29

Year: 2001

Objective of Study: To evaluate the stock price impact of restructuring announcements before

and after the Commercial Rehabilitation Law (CRL) using event analysis

Summary

After World War II, Japans corporate governance system was based on cross shareholdings and

long term relationships among a group of firms, also known as keiretsu. The leading role in this

keiretsu was played by the main bank in the group. In the long period of economic stagnation for

the last ten years, the government realized the importance of restoring profitability in the

corporate sector. This led to the formation of CRL. This research paper attempts to find the stock

price impact of the restructuring announcements by taking a sample of 1011 restructuring

announcements before and after the formation of CRL. The results of the study show a more

positive price impact during the post CRL period, compared to pre CRL period. Also, the

negative impact of labor force reduction announcements on the announcing firms stock price

during the pre CRL period disappeared in the pre CRL period.


This research paper successfully attempted to analyze the impact of restructuring announcements

on stock prices. The event considered for research is also unique. The author of this report has

also used the same statistical technique of calculating abnormal return as in other research papers

written by other authors.


30/74


30

Price Prediction Using the Box- Jenkins (BJ)

Methodology

There are different statistical tools and techniques used for price prediction on the basis of

historical price data and application of ARMA Modeling Process is one of them. In this

application I have tried to make the data series stationary, so that present data become good

predictor of future data.

Given a time series of data Xt, the ARMA model is a tool for understanding and, perhaps,

predicting future values in this series. The model consists of two parts, an autoregressive

(AR) part and a moving average (MA) part. The model is usually then referred to as the

ARMA (p,q) model where p is the order of the autoregressive part and q is the order of the

moving average part

SCOPE OF ANALYSIS

Investor uses different investment strategy, and different price matrix for identifying the right

price and appropriate timing for a particular stock purchase. Through the inclusion of ARMA

modeling I have tried to make a comparative analytical metrics, which provides investor an

opportunity to make a comparative analysis of events (through Market and Risk Adjusted

Model) as well as appropriate price (through ARMA).

Considering the broad nature of this ARMA modeling and my present understanding of the topic

the price prediction has been limited to In-Sample Prediction. This In sample prediction gives an

opportunity to understand the accuracy and aptness of present model, which can be used for Out-Sample Price Prediction.


31/74


31

OUTLINE OF ANALYSIS

The main aim of this analysis is to predict the price of stock during the event window, so that aninvestor can use this prediction, but considering the vast nature of the analysis and understanding

of the topic analysis has been limited to In- sample prediction

DEFINITION OF VARIABLES

An Autoregressive (AR) Process:

Let Ytrepresent stock price at time t. If we model Ytas

(Yt) = 1(Yt1 ) + ut

where is the mean ofYand where utis an uncorrelated random error term with zero mean and

constant variance 2 (i.e., it is white noise), then we say that Yt follows a first-order

autoregressive, or AR(1), stochastic process. Here the value ofYat time tdepends on its value

in the previous time period and a random term; the Yvalues are expressed as deviations from

their mean value. In other words, this model says that the forecast value ofYat time tis simply

some proportion (= 1) of its value at time (t 1) plus a random shock or disturbance at time t;

again the Yvalues are expressed around their mean values.

A Moving Average (MA) Process:

If we model (Stock Prices) Yas follows:

Yt= +0ut+1ut1

where is a constant and u, as before, is the white noise stochastic error term. Here Yat time tis

equal to a constant plus a moving average of the current and past error terms. Thus, in the present

case, we say that Yfollows a first-order moving average, or an MA(1), processBut ifYfollows the expression

Yt= +0ut+1ut1 +2ut2

then it is an MA(2) process.


32/74


32

An Autoregressive and Moving Average (ARMA) Process

There is high probability that Y (Stock Price) follows both AR and MA process and is therefore

ARMA. Thus, ifYtfollows an ARMA (1, 1) process, it can be written as

Yt= + 1Yt1 +0ut+1ut1

because there is one autoregressive and one moving average term. Here represents a constant

term. In general, in an ARMA (p, q) process, there will be p autoregressive and q moving

average terms.

Unit Root Test:

A unit root test tests whether a time series variable is non-stationary, using an autoregressive

model. To test the unit root we can use the Augmented Dickey-Fuller test or the Phillips-

Perron test. Both the tests use the existence of a unit root as the null hypothesis.

Dickey Fuller Test:

The Dickey-Fuller test tests whether a unit root is present in an autoregressive model. The main

objective of Dickey- Fuller test is, If the series y is (trend-) stationary, then it has a tendency to

return to a constant (or deterministically trending) mean. Therefore large values will tend to be

followed by smaller values (negative changes), and small values by larger values (positive

changes). Accordingly, the level of the series will be a significant predictor of next period's

change, and will have a negative coefficient. If, on the other hand, the series is integrated, then

positive changes and negative changes will occur with probabilities that do not depend on the

current level of the series; in a random walk, where you are now does not affect which way you

will go next.

There is also called the Augmented Dickey Fuller (ADF), which removes all the structural effect

(autocorrelation) in the time series and then tests using the same procedure.
http://en.wikipedia.org/wiki/Dickey-Fuller_testhttp://en.wikipedia.org/wiki/Dickey-Fuller_testhttp://en.wikipedia.org/w/index.php?title=Phillips-Perron_test&action=edit&redlink=1http://en.wikipedia.org/w/index.php?title=Phillips-Perron_test&action=edit&redlink=1http://en.wikipedia.org/w/index.php?title=Phillips-Perron_test&action=edit&redlink=1http://en.wikipedia.org/wiki/Unit_roothttp://en.wikipedia.org/wiki/Autoregressivehttp://en.wikipedia.org/wiki/Augmented_Dickey-Fuller_testhttp://en.wikipedia.org/wiki/Augmented_Dickey-Fuller_testhttp://en.wikipedia.org/wiki/Autoregressivehttp://en.wikipedia.org/wiki/Unit_roothttp://en.wikipedia.org/w/index.php?title=Phillips-Perron_test&action=edit&redlink=1http://en.wikipedia.org/w/index.php?title=Phillips-Perron_test&action=edit&redlink=1http://en.wikipedia.org/wiki/Dickey-Fuller_test


33/74


33

Akaike's information criterion:

Akaike's information criterion (AIC) is a measure of the goodness of fit of an estimated

statistical model. It is grounded in the concept ofentropy, in effect offering a relative measure of

the information lost when a given model is used to describe reality and can be said to describe

the tradeoff between bias and variance in model construction, or loosely speaking that of

precision and complexity of the model.

The AIC is not a test on the model in the sense of hypothesis testing, rather it is a tool for Model

selection. Given a data set, several competing models may be ranked according to their AIC,

with the one having the lowest AIC being the best. From the AIC value one may infer that e.g

the top three models are in a tie and the rest are far worse, but one should not assign a value

above which a given model is 'rejected'.

Durbin- Watson Test:

The Durbin-Watson statistic is a test statistic used to detect the presence of autocorrelation in

the residuals from a regression analysis. Its value always lies between 0 and 4. A value of 2

indicates there appears to be no autocorrelation. If the Durbin-Watson statistic is substantially

less than 2, there is evidence of positive serial correlation. As a rough rule of thumb, if Durbin-

Watson is less than 1.0, there may be cause for alarm. Small values ofdindicate successive error

terms are, on average, close in value to one another, or positively correlated. Large values of d

indicate successive error terms are, on average, much different in value to one another, or

negatively correlated.

THE BOX-JENKINS METHODOLOGYThe BJ methodology involves making the time series data stationary, but data series in itself does

not tell that whether the data follows purely AR process (and if so, what is the value ofp) or a

purely MA process(and if so, what is the value ofq) or an ARMA process (and if so, what are
http://en.wikipedia.org/wiki/Statistical_modelhttp://en.wikipedia.org/wiki/Information_entropyhttp://en.wikipedia.org/wiki/Kullback-Leibler_divergencehttp://en.wikipedia.org/wiki/Biashttp://en.wikipedia.org/wiki/Variancehttp://en.wikipedia.org/wiki/Model_selectionhttp://en.wikipedia.org/wiki/Model_selectionhttp://en.wikipedia.org/wiki/Autocorrelationhttp://en.wikipedia.org/wiki/Regression_analysishttp://en.wikipedia.org/wiki/Regression_analysishttp://en.wikipedia.org/wiki/Autocorrelationhttp://en.wikipedia.org/wiki/Model_selectionhttp://en.wikipedia.org/wiki/Model_selectionhttp://en.wikipedia.org/wiki/Model_selectionhttp://en.wikipedia.org/wiki/Variancehttp://en.wikipedia.org/wiki/Biashttp://en.wikipedia.org/wiki/Kullback-Leibler_divergencehttp://en.wikipedia.org/wiki/Information_entropyhttp://en.wikipedia.org/wiki/Statistical_model


34/74


34

the values ofp and q) or an ARIMA process, in which case we must know the values ofp, d, and

q. Thus we can divide the full methodology in four steps:

Identification of the Model:

First step of the application of The BJ methodology involves the identification of the model. The

chief tools in identification are the autocorrelation function (ACF), the partial

autocorrelation function (PACF), and the resulting correlograms, which are simply the plots

of ACFs and PACFs against the lag length.

If the analysis of data shows that time series is not stationary, we have to make it stationary

before we can apply the BoxJenkins methodology. For making the series stationary we plotted

the first differences of stock price series. A formal application of the Dickey

Fuller unit roottest shows that whether the series is stationary or not. We can also check that whether the series

has become stationary or not, we can visually check from the estimated ACF and PACF

correlograms.


35/74


35

Parameter Estimation of the Model

Now the question arises whether the stock price data follows ARMA pattern. One way of

accomplishing this is to consider the ACF and PACF and the associated correlograms of a

selected number of ARMA processes, such as AR(1), AR(2), MA(1), MA(2), ARMA(1, 1),

ARIMA(2, 2), and so on. Since each of these stochastic processes exhibits typical patterns of

ACF and PACF, if the time series under study fits one of these patterns we can identify the time

series with that process. Next, we will require applying diagnostic tests to find out if the chosen

ARMA model is reasonably accurate.

Diagnostic Checking

How do we know that the model in is a reasonable fit to the data? One simple diagnostic is to

obtain residuals and obtain the ACF and PACF of these residuals, say, up to lag 25, then we see

that whether the estimated AC and PACF is individually statistically significant or not. In other

words, the correlograms of both autocorrelation and partial autocorrelation give the impression

that the residuals estimated from the model are purely random. Hence, we will be able to

conclude that the model used for estimation is reasonably correct and we do not require using

another ARMA or ARIMA model.

Forecasting:

Forecasting represent the most important aspect of this modeling process, To check the accuracy

of the modeling the result can be computed for In sample data. In forecast E-views computes

(n-period-ahead) dynamic forecasts of an estimated equation. Forecast computes the forecast for

all observations in the current sample. As per the requirement instruction can be made to

compare the forecasted data to actual data, and to compute summary statistics.


36/74


36

Application of ARMA Modeling Process

In this report, I have checked the applicability of ARMA Model on the stock prices of Imclone

Systems Inc. (IMCL). Here, I have tried to make the data stationery so that the mean, variance

and autocovariance become constant over a period of time and in this way, the historical data

becomes a good predictor of the future data. All this application has been made using E-views

3.0.

If we take a look at the raw stock price data of IMCL (below Table 1), the correlogram shows a

high degree of auto-correlation (ACF) and partial auto-correlation (PACF). If an investor wants

to predict the future stock price on the basis of this data, the prediction will be far from being

correct. In such cases, we need to remove or minimize ACF and PACF from the data.

Table 1: Correlogram of Stock prices of IMCL before application of ARMA

Sample: 1/02/2006

4/04/2008

Included observations: 583

Autocorrelation

Partial

Correlation AC PAC Q-Stat Prob

.|******** .|******** 1 0.983 0.983 566.75 0

.|*******| .|. | 2 0.966 -0.049 1114 0

.|*******| .|. | 3 0.948 0.001 1642.3 0

.|*******| .|. | 4 0.93 -0.031 2151.3 0

.|*******| .|. | 5 0.912 0.012 2642 0

.|*******| .|. | 6 0.895 0.004 3115.2 0

.|*******| .|. | 7 0.88 0.057 3573.5 0

.|*******| .|. | 8 0.866 0.015 4018 0

.|*******| .|. | 9 0.851 -0.038 4447.9 0

.|****** | .|. | 10 0.835 -0.022 4863.1 0

.|****** | .|. | 11 0.821 0.035 5265.1 0

.|****** | .|. | 12 0.809 0.045 5655.7 0

.|****** | .|. | 13 0.797 0.004 6035.3 0

.|****** | .|. | 14 0.785 0.016 6404.8 0


37/74


37

.|****** | .|. | 15 0.772 -0.054 6763.1 0

.|****** | .|. | 16 0.758 -0.05 7108.9 0

.|****** | .|. | 17 0.744 0.011 7442.8 0

.|****** | .|. | 18 0.73 -0.013 7764.6 0

.|****** | *|. | 19 0.713 -0.073 8072.3 0

.|***** | .|. | 20 0.696 -0.033 8365.8 0

.|***** | .|. | 21 0.681 0.05 8647.1 0

.|***** | *|. | 22 0.664 -0.061 8915.1 0

.|***** | .|. | 23 0.648 0.026 9171 0

.|***** | .|. | 24 0.634 0.029 9416 0

.|***** | .|. | 25 0.619 -0.021 9650.2 0

.|***** | .|. | 26 0.604 -0.036 9873.5 0

.|**** | .|. | 27 0.588 -0.021 10086 0

To do so, we first need to check the type of the data in order to apply correct model on it. This

can be done by applying a unit root test on the data. In this case, Ive done a first difference

Augmented Dickey-Fuller test to check the unit root. This test assumes the null hypothesis that

there is a unit root in the data. Here, the absolute value of ADF test comes out to be 23.4, which

is greater than the tabulated values at 1%, 5% and 10% level of significance. Hence, we do not

accept the null hypothesis and conclude that the data doesnt follow a static pattern and is not

affected by the past data.

Table 2: Test of Unit Root at 1st

Difference

ADF Test Statistic -23.43528 1% Critical Value* -2.56925% Critical Value -1.940010% Critical Value -1.6159

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

Augmented Dickey-Fuller Test EquationDependent Variable: D(IMCLP,2)Method: Least Squares

Sample(adjusted): 1/04/2006 3/27/2008Included observations: 582 after adjusting endpoints

Variable Coefficient Std. Error t-Statistic Prob.

D(IMCLP(-1)) -0.972967 0.041517 -23.43528 0.0000


38/74


38

R-squared 0.485936 Mean dependent var -0.002194Adjusted R-squared 0.485936 S.D. dependent var 1.279607S.E. of regression 0.917456 Akaike info criterion 2.667292Sum squared resid 489.0424 Schwarz criterion 2.674795Log likelihood -775.1820 Durbin-Watson stat 1.997268

Now, by looking at the correlogram, we need to apply the correct model. If autocorrelation is

high in the data, then we need to apply MA(1) process and if there is high partial autocorrelation

in the data, then we need to apply AR(1) process. We need to check the probability values at the

rightmost column of the table below, these values should be greater than 0.05. If they are not, we

need to apply another model like AR(2), MA(2), etc.

In this case, we get the desired probabilities of greater than 0.05 after applying AR(1) MA(1)model. Now, the data has been made stationery and the future values can be forecasted on the

basis of this data (data pattern presented in Table 3 below).

Table3: Correlogram of IMCL Stock price after application of ARMA Modeling process

Sample: 1/03/2006 3/27/2008

Included observations: 583

Q-statistic probabilities adjusted for 2

ARMA term(s)

Autocorrelation

Partial

Correlation AC PAC

Q-

Stat Prob

.|. | .|. | 1 -0.003 -0.003 0.0069

.|. | .|. | 2 -0.001 -0.001 0.0081

.|. | .|. | 3 0.012 0.012 0.0951 0.758

.|. | .|. | 4 0 0 0.0951 0.954

.|. | .|. | 5 -0.019 -0.019 0.3056 0.959

.|. | .|. | 6 -0.035 -0.035 1.0364 0.904

.|. | .|. | 7 -0.01 -0.011 1.0989 0.954

.|. | .|. | 8 0.041 0.041 2.0803 0.912

.|. | .|. | 9 0.047 0.048 3.3725 0.849

.|. | .|. | 10 -0.047 -0.047 4.6692 0.792

.|. | .|. | 11 -0.042 -0.045 5.7022 0.769

.|. | .|. | 12 0.003 0 5.7087 0.839

.|. | .|. | 13 -0.028 -0.026 6.1861 0.861


39/74


39

.|. | .|. | 14 0.054 0.06 7.9023 0.793

.|. | .|. | 15 0.033 0.036 8.553 0.806

.|. | .|. | 16 0.026 0.022 8.9694 0.833

.|. | .|. | 17 -0.008 -0.018 9.005 0.877

.|* | .|* | 18 0.1 0.1 15.07 0.52

.|. | .|. | 19 0.038 0.048 15.961 0.527

.|. | .|. | 20 -0.041 -0.034 16.975 0.525

.|* | .|* | 21 0.067 0.067 19.666 0.415

.|. | .|. | 22 -0.049 -0.052 21.1 0.391

.|. | .|. | 23 -0.032 -0.04 21.706 0.417

.|. | .|. | 24 -0.027 -0.022 22.14 0.452

.|. | .|. | 25 0.028 0.042 22.626 0.483

.|. | .|. | 26 0.007 0.005 22.655 0.54

.|. | .|. | 27 0.052 0.047 24.322 0.501

.|. | .|. | 28 0.059 0.065 26.489 0.436

.|. | .|. | 29 -0.026 -0.025 26.913 0.468

.|. | .|. | 30 0.045 0.037 28.155 0.456

Table4: Table values after application of Estimation Model

Dependent Variable: IMCLPMethod: Least SquaresDate: 05/19/08 Time: 22:25Sample(adjusted): 1/03/2006 3/26/2008Included observations: 582 after adjusting endpointsConvergence achieved after 6 iterationsBackcast: 1/02/2006

Variable Coefficient Std. Error t-Statistic Prob.

C 38.50448 3.313260 11.62132 0.0000AR(1) 0.987554 0.007243 136.3507 0.0000MA(1) 0.037714 0.042169 0.894354 0.3715

R-squared 0.972693 Mean dependent var 36.81792Adjusted R-squared 0.972598 S.D. dependent var 5.524518S.E. of regression 0.914500 Akaike info criterion 2.664262

Sum squared resid 484.2235 Schwarz criterion 2.686770Log likelihood -772.3004 F-statistic 10312.01Durbin-Watson stat 1.998543 Prob(F-statistic) 0.000000

Inverted AR Roots .99Inverted MA Roots -.04


40/74


40

Figure 7: In- Sample prediction of stock Price for the period of 1 month

To check the accuracy of the model, I did an in-sample test. In this test, I checked for the values

predicted by the model which were originally supplied to it before the application of unit root

test. Here, in the graph, the blue line shows the predicted value of the stock on the given time,

and the red lines show the standard error in the prediction.


41/74


41

Table 5: Model Accuracy check for in-sample prediction

The accuracy of the model can be checked by the Theil Inequality Coefficient (Refer Table 5),

which should be closer to zero in order to make the model accurate; and Covariance Coefficient,

which should be as high as possible. In this case, the value for Theil Inequality Coefficient is

comes out to be 0.010083, which is quite closer to zero. Also, the value of Covariance

Coefficient comes out to be 0.973. So we can say that the model is accurate enough to predict the

in sample future values.

However, the actual soundness of the model can be checked only by out sample test, which could

not be carried out due to insufficiency of time and limitation of my knowledge in the concerned

field.


42/74


42

Comparative Analysis of Effect of Dividend Declaration on

Stock Return Using Risk adjusted and Single Index Model

The one of the objective of the project, as already stated, is to find out the impact of major events

and to establish the time length of impact of these events on the stock price return. Since the

event analysis does not follow any specific method and different research papers have used

different methodology to substantiate the effect of the event on stock price return, therefore in

this analysis I have used Risk Adjusted Method of analysis to substantiate my result and to make

a comparative study between Single Index and Risk adjusted model. The description regarding

the application of Single index model has been given in interim report, here the application

process of Risk Adjusted model has been explained.

DATA COLLECTION

Data is collected from three different sources. For the events data of Dividend in the years 2007

and 2008, our companys website m4dataquest.com is used, as the data is available in the easiest

format on this website. However, in cases where company doesnt have database for any quarter,

I collected this data from the official website of the company. The data for the event of Dividend

is collected using these two sources. For the data regarding the stock prices of the company,

Yahoo Finance is used. Also, to collect the corresponding data for the S&P 500 index, the same

website is used.

SAMPLE


43/74


43

For undertaking this analysis, a same sample of 78 major Dividend Declaration events was taken,

which has been used for Single Index Model. All the events were distributed across the different

sector. The sampling method has been used here is Judgmental Sampling

OUTLINE OF THE ANALYSIS

The main aim of this analysis is to establish how share prices return patterns around various

critical corporate action processing dates in both the model.

Event Window: Event window is defined as the period over which the impact of the event is

studied. During this period, the stock is expected to give a different return than normal returns in

the absence of the event. In this study, the event window is taken of -15 to +15 days of the

occurring of the event.

Estimation Window: Estimation window is defined as the period over which the normal return

of the stock is calculated. During this period, the stock is assumed to give the normal returns.

This period is taken as to be free from any sudden event or announcement. Generally, this

window is taken for as many numbers of days as possible in order to smoothen out the effect of

any unwanted or abnormal event. In this study, the estimation window is taken for a period of 30

trading days before the occurrence of the event. This window takes into account is of 30 trading

days (i.e. -60 to -30 days) of the stock. It is assumed that in such a long period, the effect of all

the events is averaged out and nullified.


44/74


44

DEFINITIONS OF VARIABLES

The empirical analysis presented in this report focuses on share price behavior around dividend

declaration dates. This section sets out definitions of the variables used in this study.

Share Price Return: Share price return is defined as the change in share price of the company

over its previous closing price. For the purpose of adjustment of risk the 10 Yr. US T- Bills rate

has been discounted from this return. It is calculated as:

((PtPt-1) / Pt-1)-Rf

METHODOLOGY

The methodology used in this study involves the use of the Risk Adjusted Model. This model

helps in finding the significance of the share price returns generated during the event window.

To begin with, let us assume a null and an alternative hypothesis,

H0:Dividend declaration does not create stock price volatility

H1:Dividend declaration creates stock price volatility

This method requires a data set of two variables, one dependent and one independent, to apply

the regression model. For this purpose, risk adjusted share price return is taken as dependent

variable and risk adjusted market return is taken as the independent variable. Applying

regression model on these variables, the values of alpha and beta are calculated.


45/74


45

Then, a regression equation is developed to calculate the expected return of the share during the

event window. The regression equation used is:

where,

- is the constant return on the share, irrespective of other factors.

is the slope of the equation, or the sensitivity of the stock return towards the market return.

Then, the risk free rate has been discounted from expected return and it is compared with the

actual adjusted T bills return and the difference of the two is termed as the abnormal return.

The variance and standard error for this abnormal return are calculated and a z-testis applied and

the standardized values are calculated using the formula,

where,

xis the average abnormal return of the sample events

- is the abnormal return of the whole population (assumed to be zero)

- is the standard error of the sample

This standardizedz-value is compared with the criticalz-value and if the standardized value is

greater than the critical value, the event is considered to have generated abnormal return on that

day.


46/74


46

When Companies Make Dividend Surprise (Abnormal

Return)

KEY FINDINGS &CONCLUSION

Comparing for the abnormal returns in both Risk Adjusted and Single Index Model, the

following results were found:

During the event window of 10 days, the standardized z-value is smaller than the critical

z-value for all the days except Day 0, Day 1, Day 2 and Day 5 at 10% level of

significance in case of Single Index Model, While the Abnormal return was significant

just on Day 0 in case of Risk Adjusted model.

In case of analysis with Single Index the null hypothesis is accepted for rest of days and it

is proved that significant difference between the share price return of estimation window

and event window last for 2 days after the event happens, but in case of analysis with

Risk adjusted model the Zcritical value shows the same pattern for the next three days but it

is not significant except on Day 0.

The abnormal return from the Single index model shows a clear pattern of return, which

start 1 day before the event and last for next 2 days, while in case of Risk Adjusted

model, it does not show any clear pattern, it start 4 days before the event and last for next

3 days after event, showing peak on event day.

It can be concluded from the above test that the event Dividend (surprise) does not

generate any significant abnormal share price returns for the full event window but last

for few days after the event.


47/74


47

Figure 8: Abnormal Return during Event Window (Using Single Index Model)

Figure 9: Abnormal Return during Event Window (Using Risk Adjusted Model)

-0.60%

-0.40%

-0.20%

0.00%

0.20%

0.40%

0.60%

0.80%

-0.40%

-0.20%

0.00%

0.20%

0.40%

0.60%

0.80%

-10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10

MeanAR


48/74


48

Figure 10: Pre and Post Event CUM Ab. Return (Single Index Model)

Figure 11: Pre and Post Event CUM Ab. Return (Risk Adjusted Model)

-0.10%

0.00%

0.10%

0.20%

0.30%

0.40%

0.50%

0.60%

0.70%

Pre Event Event Day Post Event

0.00%

0.50%

1.00%

1.50%

2.00%

2.50%

Pre Event Event date Post Event


49/74


49

Table 6: Calculation of significance of Ab. Return during event window (Single Index Model)MeanAR STD STDErr Z Score Significance Day

0.0028 0.014714 0.00171 1.663863 ***** -10-0.0013 0.015454 0.001796 -0.69594 ~ -9

-0.0003 0.012385 0.00144 -0.2151 ~ -8

0.0011 0.015177 0.001764 0.608037 ~ -7

-0.0039 0.023863 0.002774 -1.39177 ~ -6

-0.0033 0.039801 0.004627 -0.7184 ~ -5

-0.0007 0.017879 0.002078 -0.33865 ~ -4

-0.0008 0.014872 0.001729 -0.44227 ~ -3

-0.0001 0.017541 0.002039 -0.04959 ~ -2

0.0022 0.020602 0.002395 0.908352 ~ -1

0.0066 0.025804 0.003 2.202472 ***** 0

0.0060 0.01908 0.002218 2.703373 ***** 1

0.0052 0.025638 0.00298 1.735902 ***** 2

-0.0005 0.032941 0.003829 -0.13263 ~ 3

-0.0013 0.015258 0.001774 -0.73952 ~ 4

0.0038 0.019448 0.002261 1.684305 ***** 5

-0.0012 0.018039 0.002097 -0.56872 ~ 6

-0.0002 0.021545 0.002505 -0.09208 ~ 7

-0.0010 0.019981 0.002323 -0.41908 ~ 8

0.0027 0.017673 0.002054 1.304271 ~ 9

0.0005 0.018462 0.002146 0.248394 ~ 10

****z calculated is more than z tabulated at 10% significance level

~~~~z calculated is less than z tabulated at 10% significance level


50/74


50

Table 7: Calculation of significance of Ab. Return during event window (Risk Adjusted Model)

EventWindow MEAN Rt. St. DEV. St. Error Z Value

Critical Z

Value at

5% level

Critical Z

Value at

10% level

D-15 0.000209 0.016296 0.002006 0.104405 1.96 1.65

D-14 0.000785 0.017305 0.00213 0.368394 1.96 1.65

D-13 0.001016 0.015357 0.00189 0.537304 1.96 1.65

D-12 0.003721 0.022894 0.002818 1.32046 1.96 1.65

D-11 0.001571 0.022839 0.002811 0.558783 1.96 1.65

D-10 -0.0034 0.018027 0.002219 -1.53069 1.96 1.65

D-9 0.000142 0.019571 0.002409 0.059114 1.96 1.65

D-8 0.002216 0.012927 0.001591 1.392571 1.96 1.65

D-7 -0.00052 0.013815 0.001701 -0.30834 1.96 1.65

D-6 0.005797 0.026949 0.003317 1.747623 1.96 1.65

D-5 0.006 0.041559 0.005116 1.172931 1.96 1.65

D-4 0.000955 0.02044 0.002516 0.379518 1.96 1.65

D-3 0.002706 0.018651 0.002296 1.178613 1.96 1.65

D-2 0.001774 0.02072 0.002551 0.695404 1.96 1.65

D-1 0.001482 0.022894 0.002818 0.525843 1.96 1.65

D 0.00702 0.028705 0.003533 1.986792 1.96 1.65

D+1 0.003829 0.021988 0.002707 1.414757 1.96 1.65D+2 0.004829 0.026936 0.003316 1.456459 1.96 1.65

D+3 0.002147 0.016699 0.002055 1.044419 1.96 1.65

D+4 0.001502 0.019298 0.002375 0.632353 1.96 1.65

D+5 -0.00404 0.020136 0.002479 -1.63094 1.96 1.65

D+6 0.001455 0.021008 0.002586 0.562589 1.96 1.65

D+7 0.001564 0.023424 0.002883 0.542467 1.96 1.65

D+8 0.001793 0.022827 0.00281 0.638289 1.96 1.65

D+9 -0.00145 0.021949 0.002702 -0.53628 1.96 1.65

D+10 0.001032 0.024011 0.002956 0.349104 1.96 1.65

D+11 -0.00314 0.031438 0.00387 -0.81239 1.96 1.65D+12 0.000362 0.023572 0.002901 0.124919 1.96 1.65

D+13 -0.00233 0.023705 0.002918 -0.79971 1.96 1.65

D+14 0.000612 0.019891 0.002448 0.250008 1.96 1.65

D+15 0.003432 0.025013 0.003079 1.114552 1.96 1.65


51/74


51

When Companies Make Dividend Surprise (Increases dividend

between 0 to 10%)


Comparing for the abnormal returns in both Risk Adjusted and Single Index Model for 0 t o10%

dividend increase cases, the following results were found:


z-value for all the days except Day 2 at 10% level of significance in case of Single Index

Model, while the Abnormal return was significant on most of days except on event day in

the 10 days event window in case of Risk Adjusted model.



and event window except on D+2, but in case of analysis with Risk adjusted model the

Zcritical value shows significant on most of days except on Day 0.



model, it does not show any clear pattern, rather in most of the cases it does not give anyabnormal return or may say gives negative abnormal return.

Even though both the models show a different pattern but It can be concluded from the

investors point of view that the event Dividend (surprise of 10 to 20%) does not generate

any significant abnormal share price returns for the full event window and is not of much

economic significance for the investor.


52/74


52

Figure 13: Abnormal Return during event window through Single Index model

Figure 13: Abnormal Return during event window through risk adjusted model

Mean AR

-0.60%

-0.40%

-0.20%

0.00%

0.20%

0.40%

0.60%

0.80%

-10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10

Mean AR

-3.00%

-2.00%

-1.00%

0.00%

1.00%

2.00%

3.00%

4.00%


53/74


53

Figure 15: Cumulative Abnormal Return during event window through Single Index model

Figure 15: Cumulative Abnormal Return during event window through risk adjusted model


-0.50%

0.00%

0.50%

1.00%

1.50%

2.00%

2.50%


-1.00%

0.00%

1.00%

2.00%

3.00%

4.00%

5.00%



54/74


54

Table 8: Calculation of Significance ofAbnormal Return during event window (Using Single Index

Model)

MeanAR STD STDErr Z Score Significance Day

0.0041 0.011542 0.002356 1.755496 ***** -10

-0.0019 0.010728 0.00219 -0.87397 ~ -9

-0.0006 0.009583 0.001956 -0.31526 ~ -8

-0.0010 0.010466 0.002136 -0.46351 ~ -7

-0.0013 0.014771 0.003015 -0.42564 ~ -6

-0.0017 0.019646 0.00401 -0.43223 ~ -5

-0.0001 0.012978 0.002649 -0.03985 ~ -4

-0.0014 0.013772 0.002811 -0.48862 ~ -3

0.0007 0.015306 0.003124 0.233387 ~ -2-0.0001 0.013679 0.002792 -0.04626 ~ -1

0.0054 0.017329 0.003537 1.515998 ~ 0

0.0044 0.015641 0.003193 1.365998 ~ 1

0.0062 0.015818 0.003229 1.923241 ***** 2

0.0045 0.015607 0.003186 1.42605 ~ 3

-0.0002 0.015584 0.003181 -0.05763 ~ 4

0.0036 0.017353 0.003542 1.025322 ~ 5

-0.0024 0.01893 0.003864 -0.62542 ~ 6

-0.0051 0.019655 0.004012 -1.27742 ~ 7

-0.0013 0.020229 0.004129 -0.32661 ~ 8

0.0062 0.017812 0.003636 1.698365 ***** 9

0.0048 0.013725 0.002802 1.706382 ***** 10




55/74


55

Table9: Calculation of significance ofAbnormal Return during event window (Using Risk Adjusted

Model)


Critical Z

Value at

5% level

Critical Z

Value at

10% level

D-15 -0.00339 0.01075 0.001323 -2.5645 1.96 1.65

D-14 0.010437 0.019016 0.002341 4.458839 1.96 1.65

D-13 0.009716 0.014693 0.001809 5.371938 1.96 1.65

D-12 -0.0059 0.033515 0.004125 -1.43042 1.96 1.65

D-11 0.009379 0.031449 0.003871 2.422849 1.96 1.65

D-10 0.002794 0.0150860.001857 1.504635 1.96 1.65

D-9 -0.0007 0.016203 0.001994 -0.34948 1.96 1.65

D-8 0.003732 0.011037 0.001359 2.74719 1.96 1.65

D-7 0.001109 0.01351 0.001663 0.666874 1.96 1.65

D-6 0.004979 0.021028 0.002588 1.923679 1.96 1.65

D-5 0.006638 0.021048 0.002591 2.562182 1.96 1.65

D-4 0.014477 0.016364 0.002014 7.187021 1.96 1.65

D-3 -0.00535 0.019098 0.002351 -2.27466 1.96 1.65

D-2 -0.00215 0.019242 0.002369 -0.90587 1.96 1.65

D-1 0.003151 0.016568 0.002039 1.545291 1.96 1.65

D -0.00221 0.021112 0.002599 -0.85003 1.96 1.65D+1 -0.00122 0.020598 0.002535 -0.48253 1.96 1.65

D+2 0.004342 0.016217 0.001996 2.175177 1.96 1.65

D+3 -0.01113 0.016684 0.002054 -5.41973 1.96 1.65

D+4 -0.00708 0.015484 0.001906 -3.71351 1.96 1.65

D+5 0.00939 0.016882 0.002078 4.518755 1.96 1.65

D+6 0.001785 0.019164 0.002359 0.756822 1.96 1.65

D+7 0.002815 0.023008 0.002832 0.993865 1.96 1.65

D+8 0.008657 0.025163 0.003097 2.794813 1.96 1.65

D+9 -0.0208 0.019778 0.002435 -8.54203 1.96 1.65

D+10 -0.01991 0.02513 0.003093 -6.43779 1.96 1.65D+11 0.014442 0.020321 0.002501 5.773619 1.96 1.65

D+12 -0.00056 0.020558 0.002531 -0.22295 1.96 1.65

D+13 0.015192 0.025429 0.00313 4.85357 1.96 1.65

D+14 0.013194 0.01616 0.001989 6.632891 1.96 1.65

D+15 0.033728 0.021404 0.002635 12.80176 1.96 1.65


56/74


56


between 10 to 20%)


Comparing for the abnormal returns in both Risk Adjusted and Single Index Model for 10 to

20% dividend increase, the following results were found:


z-value for all the days except Day -5 at 10% level of significance in case of Single IndexModel, while the same kind of result was also presented by Risk Adjusted model. The

significance result at Day -5 also does not lead to any conclusion that whether this is due

to the effect of event itself.



and event window exist on D-5 days, but in case of analysis with Risk adjusted model the

Zcritical value shows the same pattern and is significant on D -5, D -3.


start 1 day before the event and remain positive just on event day, while in case of Risk

Adjusted model, it does not show any clear pattern, event on day 0 it shows negligible

abnormal return..

It can be concluded from the above test that the event Dividend (surprise of 10 to 20%)

does not generate any significant abnormal share price returns for the full event window.


57/74


57

Figure 16: Abnormal return during event window (Single Index Model)

Figure 17: Abnormal return during event window (Risk Adjusted model)

-0.60%

-0.40%

-0.20%

0.00%

0.20%

0.40%

0.60%

0.80%

-10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10

Mean AR

Mean AR

-1.00%

-0.50%

0.00%

0.50%

1.00%

1.50%


58/74


58

Figure 18: Cumulative Abnormal return during event window (Single Index Model)

Figure 19: Cumulative Abnormal return during event window (Risk Adjusted model)

-0.10%

0.00%

0.10%

0.20%

0.30%

0.40%

0.50%

0.60%

0.70%


-0.50%

0.00%

0.50%

1.00%

1.50%

2.00%

2.50%

3.00%

3.50%



59/74


59

Table 10: Calculation of significance of Abnormal Return (Single Index Model)


0.0079 0.013351 0.003238 1.885634 ***** -10

0.0022 0.009359 0.00227 0.181403 ~ -9

0.0009 0.006484 0.001573 -0.55735 ~ -8

0.0072 0.01995 0.004839 1.112362 ~ -7

-0.0002 0.010417 0.002526 -0.80561 ~ -6

-0.0038 0.010836 0.002628 -2.14654 ***** -5

0.0038 0.017802 0.004318 0.461844 ~ -4

-0.0022 0.012733 0.003088 -1.27997 ~ -3

-0.0001 0.013425 0.003256 -0.58897 ~ -2

0.0056 0.018685 0.004532 0.835932 ~ -1

0.0057 0.019998 0.00485 0.810165 ~ 0-0.0028 0.020054 0.004864 -0.94818 ~ 1

-0.0007 0.015354 0.003724 -0.68396 ~ 2

0.0010 0.015925 0.003862 -0.2056 ~ 3

-0.0023 0.019348 0.004693 -0.86368 ~ 4

-0.0008 0.01245 0.00302 -0.86496 ~ 5

0.0011 0.014512 0.00352 -0.18718 ~ 6

0.0004 0.016511 0.004004 -0.34961 ~ 7

0.0034 0.010154 0.002463 0.644895 ~ 8

0.0047 0.012255 0.002972 0.991512 ~ 9

0.0071 0.021463 0.005206 1.015864 ~ 10




60/74


60

Table 11: Calculation of significance of Abnormal Return (Single Index Model)


Critical ZValue at

5% level

Critical ZValue at

10% level

D-15 0.005467 0.013755 0.001693 3.229122 1.96 1.65

D-14 0.006012 0.011738 0.001445 4.161157 1.96 1.65

D-13 0.005644 0.014803 0.001822 3.097647 1.96 1.65

D-12 0.001859 0.0161 0.001982 0.938151 1.96 1.65

D-11 0.001001 0.016285 0.002005 0.499352 1.96 1.65

D-10 -0.00325 0.016001 0.00197 -1.64853 1.96 1.65

D-9 -0.00076 0.01429 0.001759 -0.43012 1.96 1.65D-8 0.000394 0.011054 0.001361 0.289873 1.96 1.65

D-7 -0.00203 0.011726 0.001443 -1.4046 1.96 1.65

D-6 0.002947 0.014783 0.00182 1.619598 1.96 1.65

D-5 0.004559 0.013567 0.00167 2.729996 1.96 1.65

D-4 0.001775 0.026728 0.00329 0.539479 1.96 1.65

D-3 0.007041 0.019366 0.002384 2.953758 1.96 1.65

D-2 0.003149 0.01555 0.001914 1.645272 1.96 1.65

D-1 -0.00165 0.020166 0.002482 -0.66587 1.96 1.65

D -0.00204 0.019362 0.002383 -0.85504 1.96 1.65

D+1 0.003833 0.020403 0.002511 1.526113 1.96 1.65D+2 0.001457 0.014783 0.00182 0.800791 1.96 1.65

D+3 0.001306 0.013121 0.001615 0.808758 1.96 1.65

D+4 0.003924 0.023021 0.002834 1.384844 1.96 1.65

D+5 0.000342 0.013196 0.001624 0.210315 1.96 1.65

D+6 0.003155 0.019375 0.002385 1.322976 1.96 1.65

D+7 0.000854 0.018829 0.002318 0.368359 1.96 1.65

D+8 -0.00085 0.014448 0.001778 -0.48031 1.96 1.65

D+9 -0.00216 0.012662 0.001559 -1.38698 1.96 1.65

D+10 -0.00751 0.027414 0.003374 -2.22634 1.96 1.65

D+11 0.000806 0.018328 0.002256 0.357208 1.96 1.65D+12 0.013408 0.022374 0.002754 4.868602 1.96 1.65

D+13 0.000251 0.022077 0.002718 0.092182 1.96 1.65

D+14 0.007162 0.021888 0.002694 2.658443 1.96 1.65

D+15 0.002567 0.018711 0.002303 1.114559 1.96 1.65


61/74


61


between 20 to 50%)


Comparing for the abnormal returns in both Risk Adjusted and Single Index Model for 20 to

50% dividend increase case, the following results were found:


z-value for all the days except Day-9, Day-8, Day-2 and Day 10 at 10% level ofsignificance in case of Single Index Model, While the Abnormal return was significant

just on Day 0, Day 1and day 5, 8, 10 in case of Risk Adjusted model.



and event window occurs 2 days before the event happens, but in case of analysis with

Risk adjusted model the Zcritical value shows a substantial significant value on Day0, Day

1.



model, it shows the same pattern, except on Day 3 where the abnormal return dips in this

model.

It can be concluded from the above test that even the Dividend (surprise of 20 to 50%) is

not able to generate significant abnormal share price returns for a continuous period of

time.


62/74


62

Figure 20: Abnormal Return during event window (Single Index Model)

Figure 21: Abnormal Return during event window (Risk Adjusted Model)

MeanAR

-0.80%

-0.60%-0.40%

-0.20%

0.00%

0.20%

0.40%

0.60%

0.80%

1.00%

1.20%

-10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10

MeanAR

-1.00%

-0.50%

0.00%

0.50%

1.00%

1.50%

D-15

D-13

D-11

D-9

D-7

D-5

D-3

D-1

D+1

D+3

D+5

D+7

D+9

D+11

D+13

D+15


63/74


63

Figure 22: Cum Abnormal return during event window (Single Index Model)

Figure 23: Cum Abnormal return during event window (Risk Adjusted Model)

-0.20%

0.00%

0.20%

0.40%

0.60%

0.80%

1.00%

1.20%


-0.20%

0.00%

0.20%

0.40%

0.60%

0.80%

1.00%

1.20%

1.40%

1.60%

1.80%



64/74


64

Table 12: Calculation of significance of abnormal return during event window (Single Index

model)


0.0006 0.008243 0.001999 0.10298 ~ -10

-0.0060 0.015845 0.003843 -1.67148 ***** -9

-0.0062 0.01351 0.003277 -2.02136 ***** -8

-0.0038 0.013793 0.003345 -1.25022 ~ -7

-0.0032 0.024881 0.006035 -0.59656 ~ -6

-0.0013 0.008231 0.001996 -0.82799 ~ -5

0.0022 0.013052 0.003165 0.568633 ~ -4

0.0037 0.012507 0.003033 1.080158 ~ -3

0.0075 0.013616 0.003302 2.144603 ***** -2

-0.0023 0.016999 0.004123 -0.65632 ~ -1

0.0101 0.025137 0.006097 1.588134 ~ 0

0.0058 0.015909 0.003858 1.404088 ~ 1

0.0043 0.014505 0.003518 1.108601 ~ 2

-0.0008 0.012896 0.003128 -0.39871 ~ 3

-0.0003 0.014017 0.0034 -0.19899 ~ 4

0.0036 0.012441 0.003017 1.068311 ~ 5

-0.0016 0.008932 0.002166 -0.9341 ~ 6-0.0009 0.01167 0.00283 -0.45722 ~ 7

-0.0029 0.016489 0.003999 -0.81489 ~ 8

0.0038 0.01305 0.003165 1.066799 ~ 9

.-0.0047 0.012446 0.003019 -1.69345 ***** 10




65/74


65

Table 13: Calculation of significance of abnormal return during event window (Risk Adjusted

Model)


Critical Z

Value at

5% level

Critical Z

Value at

10% level

D-15 -0.00605 0.019784 0.002435 -2.4836 1.96 1.65

D-14 -0.00187 0.018091 0.002227 -0.83851 1.96 1.65

D-13 -0.00239 0.016992 0.002092 -1.14509 1.96 1.65

D-12 0.005322 0.016005 0.00197 2.701443 1.96 1.65

D-11 -0.00308 0.017689 0.002177 -1.414 1.96 1.65

D-10 -0.00384 0.0233790.002878 -1.33577 1.96 1.65

D-9 -0.00102 0.025016 0.003079 -0.33233 1.96 1.65

D-8 0.004032 0.015871 0.001954 2.064015 1.96 1.65

D-7 0.001463 0.016449 0.002025 0.722718 1.96 1.65

D-6 0.007569 0.023617 0.002907 2.603586 1.96 1.65

D-5 0.006364 0.063322 0.007794 0.816537 1.96 1.65

D-4 -0.00063 0.016788 0.002067 -0.30309 1.96 1.65

D-3 -0.00098 0.016853 0.002075 -0.47307 1.96 1.65

D-2 -0.00427 0.018967 0.002335 -1.82794 1.96 1.65

D-1 -0.00154 0.028946 0.003563 -0.43176 1.96 1.65

D 0.01496 0.033988 0.004184 3.575823 1.96 1.65D+1 0.00953 0.0196 0.002413 3.950173 1.96 1.65

D+2 -0.00767 0.039354 0.004844 -1.5838 1.96 1.65

D+3 -0.0017 0.019065 0.002347 -0.72503 1.96 1.65

D+4 0.001381 0.020157 0.002481 0.556579 1.96 1.65

D+5 -0.00694 0.021827 0.002687 -2.58273 1.96 1.65

D+6 0.003916 0.021423 0.002637 1.485215 1.96 1.65

D+7 -0.00239 0.022385 0.002755 -0.86758 1.96 1.65

D+8 0.009786 0.017566 0.002162 4.526152 1.96 1.65

D+9 0.003557 0.028135 0.003463 1.026969 1.96 1.65

D+10 0.00713 0.015481 0.001906 3.741774 1.96 1.65D+11 0.000288 0.02544 0.003131 0.092068 1.96 1.65

D+12 -0.00746 0.023833 0.002934 -2.54231 1.96 1.65

D+13 -0.00369 0.020777 0.002557 -1.44093 1.96 1.65

D+14 0.000638 0.020832 0.002564 0.248802 1.96 1.65

D+15 0.010076 0.028071 0.003455 2.916102 1.96 1.65


66/74


66

Cross Sectional Regression Analysis

The one of the objective of the project, as already stated, is to find out the impact of major events

and to establish the time length of impact of these events on the stock price and trading volume.

To substantiate this objective and to find out that whether the investor use to be just an event

driven investor or he/she considers the other financials of the company before acting on a

particular event and investing in a particular scrip this analysis has been undertaken.

SCOPE OF THE ANALYSIS

There can be the different financial factors which an investor can consider before inve

Sip+Report +Metrics4+Analytics

Documents