# Stock Price Reaction To Annual Earnings Announcements

**Disclaimer:** This dissertation has been submitted by a student. This is not an example of the work written by our professional dissertation writers. You can view samples of our professional work here.

Any opinions, findings, conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of UK Essays.

Any decision carried out by the management of any organization needs adequate, accurate and precise information, on the basis of that information the management procures their analysis and undertake decision. If decision to be taken involves any financial aspect, this increases the scope and accuracy of the information. Financial decisions require adequate and accurate information; therefore, it is important that the behaviour of individual market is investigated for informed financial decision making, Oguzusy and Guiven (2003).

In this respect many theories were presented. One of them is about the market efficiency which is termed as efficient market hypothesis (EMH). The concept of market efficiency had been anticipated at the beginning of the century by Bachelier (1900) in his dissertation. Fama (1970) classified market efficiency in three categories namely, weak form, semi strong form and strong form of efficiency; weak form of efficiency which defines as one can't earn abnormal return by doing technical analysis of the market or of a particular stock. Technical analysis means predicting future prices by studying historical prices of a particular share or a market. The Second form of efficient market hypothesis (EMH) is semi-strong form of efficiency. This form of market efficiency makes impossible for an investor to earn extra return on security by knowing the publicly available information; this includes company's financial results, any particular event or news which affects the company the share prices adjust rapidly with these new publicly available information therefore excess return can't be earn by trading on that information. The last form of efficient market hypothesis (EMH) is the strong form of efficiency and can be define as share prices reflects all public and private information (insider information) and consequently it is not possible for a stock holder to earn extra return on the basis of these information.

According to efficient market hypothesis (EMH) the stock prices in an efficient market fully reflect their investment value Ajayi, Mehdian Perry (2004). The security pricing process instantaneously impound the available information in an efficient market and it is not possible to beat an efficient market that by using data mining, trading strategy or by any technical analysis to get consistently abnormal returns.

Efficient market hypothesis (EMH) assumed that

(1) All investors have cost-less access to currently available information about the future.

(2) They are good analysts; and

(3) They pay close attention to the market process and adjust their holdings appropriately.

Many models including Augmented Dickey Fuller (ADF) unit root test, variance ratio test (VR), Ljung Box Q-statistics, and Durbin Watson‘d' statistics have been based on this concept of informational efficiency of capital markets. However the late seventies and the eighties brought in evidences questioning the validity and highlighting various anomalies related to the Stock market efficiency. There are many focused studies that demonstrate the possible trading strategies yielding abnormal rates of return using the historical data and publicly available information ruling out the efficacy of markets. The empirical studies evidencing the inefficiency are broadly related to the following:

(1) The low price-earning (P/E) effect: Researches show that stocks with low price earning (P/E) ratios earned more for investors, which is contradictory to Efficient Market Hypothesis (EMH). Fama and French (1995) found that market and size factors in earnings help explain market and size factors in returns.

(2) The small firm and neglected firm effects: Banz (1981), Reinganum (1981) and other researchers show the size or small-firm effect in stock return. Their analysis support the evidence that small firm with low capitalization can earn higher returns than the large firm with large capitalization.

(3) Market over and under reaction: DeBondt and Thaler (1985, 1987) present evidence that is consistent with stock prices over reacting to current changes in earnings. They report positive (negative) estimated abnormal stock returns for portfolios that previously generated inferior (superior) stock price and earning performance. This could be construed as the prior period stock price behaviour over reacting to earnings developments (Bernard, 1993).

(4) The January effect: The January effect in stock returns was documented by many researchers. Their analysis suggested that January has a highest return as compared to other months. January effect was first discovered by Rozeff and Kinney (1976) for US stock markets. Later other researchers like Gultekin and Gultekin (1983), Chang and Pinegar (1986) documented the same result for other countries stock markets.

(5) The week day effect: This refers to the observation that stocks return are not independent of the day of the week effect. A notable anomaly is the Monday effect in daily stock returns, which suggests that stock returns are significantly lower or negative on Mondays relative to other week days. This ‘Monday effect' has been extensively examined not only in U.S. asset markets but in international markets as well, for example French (1980), Lakonishok and Levi (1982), Mehdian & Perry (2001) and Lakonishok & Smidt (1988).

In week day effect the last trading day that is Friday was characterized with a positive return and the first trading day that is Monday is characterized with a low or negative return. Later this interesting study was also carried out on other countries stock markets and the researchers found out the same result, but still few studies has been done on emerging Asian stock markets.

### Karachi Stock Exchange (Kse)

The Karachi Stock Exchange abbreviated as KSE is a stock exchange based in Karachi, Pakistan. It was founded in 1947 and is country's largest and oldest stock exchange, with both Pakistani and overseas listings. It is also the second oldest stock exchange in South Asia. From its inception in 1947, it has done an amazing progress. In 1950s, only 05 companies listed and 90 members were there on the exchange and at the end of 2007 the number of listed companies increased by 666 which make a total of 671 listed companies and the member on the exchange goes up from 90 to 200 during these years. Its current premises are situated in the heart of Karachi's Business District, on Stock Exchange Road.

### History

KSE is the biggest and most liquid exchange. It was recognized worldwide for performing well in 2002 by 'Business Week' magazine. US newspaper, USA Today, termed Karachi Stock Exchange (KSE) as one of the best performing bourses in the world. As of December 20, 2007, 671 companies were listed with the market capitalization of Rs.4364.312 billion (US$ 73 Billion) having listed capital of Rs.717.3 billion (US$ 12 billion). In the same year, the KSE 100 Index reached its ever highest value and closed at 14,814.85 points.

### Trading Time

The trading hours are from 9:45am to 2:15pm on weekdays and 9:30am to 1:30pm on Friday.

### Growth

The beginning of the exchange was very low with an index of 50 shares only. As the market grew, a delegate index was needed. On November 1st, 1991 the KSE-100 index was introduced and till present it is the most generally accepted measure of the exchange.

The need to reconfirm for all share indexes was felt in 1995 and to provide the beginning of index trading in future. And this was achieved on 29th of August, 1995, constructing all share indexes and introduced on 18th of September, 1995. Foreign interests were very active on KSE in 2006 and the interest continued in 2007 also. According to the estimates given by State Bank of Pakistan, foreign investment in capital markets total about US$523 Million. According to a research analyst in Pakistan, around 20% of the total free float in KSE-30 Index is held by foreign participants. There is a plan to build high rise building for the KSE as a new direction to future investments. The decision was taken by the board of directors, Karachi stock exchange (KSE). Disputes between investors and members of the Exchange are resolved through deliberations of the Arbitration Committee of the Exchange.

### Kse – 100 Index

Karachi Stock Exchange 100 Index (KSE-100 Index) is a benchmark and stock index used to compare prices overtime. In determining representative companies to compute the index, companies with the highest market capitalization are selected. To ensure full market representation, the company with the highest market capitalization from each sector is also included. The list of 100 companies listed in Karachi Stock Exchange is presented in Table # 01.

The Karachi Stock Exchange (KSE) has also launched the KSE-30 Index with base value of 10,000 points, implemented from September 1, 2006. The main feature of this index is that it based only on the free-float of shares, rather than on the basis of paid-up capital which differ it from the other indices. Unlike the Karachi Stock Exchange (KSE) which represents total return of the market, KSE-30 index is adjusted for dividends and right shares. That is, when a company announces a dividend, the other indices at Karachi Stock Exchange (KSE) are not reduced for that amount of dividend. Whereas KSE-30 Index is adjusted for dividends and right shares only

Table # 01

List of 100 Companies listed In Karachi Stock Exchange – 100 Index |
|||

No. |
Company Name |
No. |
Company Name |

1 |
Pakistan Refinery |
51 |
Pakistan Telecom. Co.Ltd |

2 |
EFU General Ins |
52 |
Sui North Gas |

3 |
Pakistan Reinsur |
53 |
New Jubilee Insurance |

4 |
EFU Life Assurance |
54 |
Mybank Limited |

5 |
Dawood Herc. |
55 |
WorldCall Telecom |

6 |
Ist.Capital Securities |
56 |
D.G.Khan Cement |

7 |
Mari Gas |
57 |
Pakistan State Oil |

8 |
Siemens Pakistan |
58 |
PICIC Growth |

9 |
Bata (Pakistan) |
59 |
Fauji Cement |

10 |
Adamjee Insurance |
60 |
Standard Chartard Bank |

11 |
Attock Refinery |
61 |
IGI Insurance |

12 |
Jahangir Siddiqque Co. |
62 |
Sui South Gas |

13 |
Pak.National Shipping Corp. |
63 |
Karachi Electric Supply Corp. |

14 |
Bank Al-Falah |
64 |
Shell Pakistan |

15 |
Meezan Bank |
65 |
Wazir Ali |

16 |
Bannu Woollen |
66 |
Samin Textiles |

17 |
JS Global Cap. |
67 |
Bestway Cement |

18 |
Rafhan Maize |
68 |
Maple Leaf Cement |

19 |
Habib Metro Bank |
69 |
Pioneer Cement |

20 |
Nestle Pakistan |
70 |
Javedan Cement |

21 |
Pakistan Elektron |
71 |
Fazal Textile |

22 |
Lucky Cement |
72 |
Pak.PTA Ltd. |

23 |
Pakistan Tobacco |
73 |
ABN AMRO Bank |

24 |
MCB Bank |
74 |
NIB Bank |

25 |
Bank AL-Habib |
75 |
Bosicor Pakistan |

26 |
Pakistan Petroleum |
76 |
Saudi Pak Bank |

27 |
Attock Petroleum |
77 |
Pakistan Cement |

28 |
Engro Chemical |
78 |
Agriautos Industries |

29 |
National Refinery |
79 |
AL-Ghazi Tractors |

30 |
ICI Pakistan |
80 |
Allied Bank |

31 |
Colgate Palmolive |
81 |
Arif Habib Securities |

32 |
Abbott (Lab) |
82 |
Askari Bank |

33 |
Habib Bank Ltd |
83 |
Atlas Honda |

34 |
Attock Cement |
84 |
Kot Addu Power Company |

35 |
Azgard Nine |
85 |
Lakson Tobacco |

36 |
Bank of Punjab |
86 |
National Bank of Pakistan |

37 |
Fauji Fertilizers Bin |
87 |
Nishat Mills |

38 |
Fauji Fertiliz |
88 |
Oil and Gas Development |

39 |
Faysal Bank |
89 |
Orix Leasing |

40 |
Ghani Glass |
90 |
Pakistan International Airlines |

41 |
GlaxoSmith |
91 |
Packages Limited |

42 |
Habib Modarba |
92 |
Pak Oilfields |

43 |
Habib Sugar |
93 |
Pak Services |

44 |
Hub Power |
94 |
Pak Suzuki |

45 |
Ibrahim Fibres |
95 |
Pakistan Intn`l Container Ter. |

46 |
Indus Motor |
96 |
Soneri Bank |

47 |
International Industries limited |
97 |
Thal Limited |

48 |
JS Investment |
98 |
UniLever Pakistan |

49 |
Kohinoor Energy |
99 |
Unilever Foods |

50 |
Cresent Commercial Bank |
100 |
United Bank |

(Source: Karachi Stock Exchange)

### History

The index was launched in late 1991 with a base of 1,000 points. By 2001, it had grown to 1,770 points. By 2005, it had skyrocketed to 9,989 points. It then reached a peak of 12,285 in February 2007. KSE-100 index touched the highest ever benchmark of 14,814 points on December 26, 2007. The graph of last 10 years of KSE growth and index points is shown. The graph clearly shows the progress and continuous increment.

Free Float Index: In order to introduce a free float index that is representative of the market, the KSE- 30 Sensitive Index was implemented with effect from September 1, 2006. The need for a market representative free float index was long felt as the capitalization weighted KSE 100 Index strongly tilted to a few scripts. Free float is based on the proportion of shares readily available for trading to the total shares issued and excludes the locked in shares. The criterion for the selection of scripts on KSE-30 index was revised on 15 February 2007 in line with international best practices to include the impact cost as a measure to gauge the liquidity of scrip.

This study is about testing the semi-strong form of Efficient Market Hypothesis (EMH) on the annual earnings announcement for the selected companies, listed on Karachi Stock Exchange (KSE) by using event study methodology (Fama et al. 1969; and Brown and Warner 1980, 1985).; Following this chapter the study is divided into six more chapters, they are;

(1) Chapter two includes detailed Research aims and objectives, it also comprises of main problem and their sub problems; hypotheses of the study are also being discussed in this chapter.

(2) In the third chapter, Review of relevant theoretical and empirical research has been done. In this chapter we have concluded that what has been done so far in this area of study both theoretically and empirically.

(3) Fourth chapter covers Research methodology, data sources and method of sampling for the data. Methodology includes formulae and tests which are being used to test semi-strong form of efficient market hypothesis (EMH) on Karachi Stock Exchange (KSE).

(4) Fifth chapter includes Research results and/or findings with supporting evidence.

(5) Sixth chapter includes the research conclusions.

(6) The seventh and the last chapter comprise of Recommendations; made with the help of Research results and/or findings.

### Scope And Limitation Of The Study

The material in this dissertation to the best of my knowledge do not contain any previously published or written documents by another person except where due acknowledgement is made in the research itself.

If any errors found in the calculations made for this research that will be the sole responsibility of the writer.

Statement Of Ethics And Originality

Due to time constraint and non availability of the company's earnings announcement data from the Karachi stock exchange web site before 2004 the study is being carried out for just three years which includes 2005, 2006 and 2007.

Moreover during the period of study which is year 2005, 2006 and 2007 there are few companies eliminated due to the non availability of the required data to carry out the calculations.

Due to the limited availability of econometrics experts for guidance irrespective of the new sophisticated models for event studies, conventional models were used in this study despite the fact they have less predictive power than the other latest models.

Aims, Objectives And Hypothesis Of The Study

The following are the Aims & Objectives of the study:

- To check whether the Semi-Strong form of Efficient Market Hypothesis (EMH) is valid for Karachi Stock Exchange` 100 – 100 Index (KSE – 100 Index).
- To examine the stock market reaction (KSE) to Annual Earnings Announcements.

### Problems

The research is comprises with one main problem which is further divided into three sub problems each problem has its own hypothesis and to be solved separately.

### Main Problem

Test whether semi-strong form of efficiency exists on Karachi Stock Exchange (KSE) or not.

### Sub Problem – One

Whether the annual earnings announcement affect complete on the day of announcement?

We will calculate the normal return and the expected return and if it is close to zero; we will say that the annual earnings announcement affect complete on the day of announcement

### Sub Problem – Two

Share holders could not earn extra return; before, and after the announcement.

We would first take the average of abnormal return and then cumulate the average abnormal return.

In case where the AARs and the CAARs are closed to zero we will conclude our results that, investor or the share holder are not able to earn abnormal return by trading on event which is earnings announcement.

Sub Problem – Three

The Average Abnormal Returns (AARs) are random.

We used Runs test to analyze the randomness in the behavior of Average Abnormal Returns (AARs). To check whether the average abnormal returns occur by chance or not, we carried out Runs test. In case where the observed numbers of runs are significantly different from the expected number of runs, we will conclude our finding as Average Abnormal Returns (AARs) do not occur randomly. Alternatively, if these results were not statistically significant, we say that Average Abnormal Returns (AARs) do occur randomly. We carried out runs test on Average Abnormal Returns (AARs) before and after the event day and also for the event window.

Hypothesis

Since the study empirically examine the Karachi Stock Exchange`s 100 Index reaction to Annual Earnings Announcement and the hypothesis being tested are:

### Hypothesis For Sub Problem One

HO: Our null hypothesis for sub problem – one is that the stock prices reactions in response to the annual earnings announcement complete on the announcement day in addition to that, abnormal returns can`t be earn by the investors on stocks by trading on stocks after the announcement day.

HO: Rit = AR = 0

H1: Rit = AR

For testing above hypothesis we compute the estimated return for the event window and then compare it to the actual return, the estimated return will be calculated by using following equation;

E (Rit) = αi + βi Rmt

Under the null hypothesis if the estimated return of a stock is closed to zero we will accept the null hypothesis and if it is not than we will reject our hypothesis and bring to a close; that announcement do affect on returns.

### Hypothesis For Sub Problem Two

HO: Our null hypothesis for sub problem – two is that returns are close to zero for average abnormal returns and their respective cumulative average abnormal returns for the selective securities in the study

HO: AAR ≈ CAAR = 0

H1: AAR ≈ CAAR

To test the above hypothesis first we will calculate the average abnormal return (AAR) and then cumulative average abnormal return (CAAR) with the help of the following formulae;

For Average Abnormal Return

Σ ARit

AAR it = i=1 .

N

Where,

i = the number of securities in the study;

N = total number of securities.

t = the days surrounding the event-day

For Cumulative Average Abnormal Return

K

CAARt = Σ AARit Where, t = -30,...0, ... +30.

t = -30

If the average abnormal return and the cumulative average abnormal return are close to zero than we accept our null hypothesis otherwise we will reject it.

2.2.3 Hypothesis for Sub Problem – Three

HO: Our null hypothesis for sub problem – three is that the difference between the no. of positive and negative average abnormal returns as not significant and they occur randomly.

HO: Z = 0

H1: Z

The null hypothesis of the test is that the observed series is a random series. A run is defined by Gibbons (1985), as

“A succession of identical symbols which are followed or preceded by different symbols or no symbol at all”

The run test is another approach to test and detect statistical dependencies (randomness). The number of runs is computed as a sequence of the price changes of the same sign (such as; + +, - - , 0 0).

When the expected number of run is significantly different from the observed number of runs, the test rejects the null hypothesis that the daily returns are random. The run test converts the total number of runs into a Z statistic. For large samples the Z statistics gives the probability of difference between the actual and expected number of runs. The Z value is greater than or equal to + 1.96, reject the null hypothesis at 5% level of significance (Sharma and Kennedy, 1977).

### Literature Review

There have been a lot of studies conducted on Efficient Market Hypothesis (EMH), a concept; developed by Fama (1960) and divided capital market into three parts on the basis of its efficiency namely weak, semi-strong and strong form. For the event study, which is linked with semi – strong form of market efficiency; below first we discuss the theoretical foundations and after that, the empirical evidence.

Theoretical Foundations

The origins of the Efficient Market Hypothesis (EMH) can be traced back to the work of two individuals, Eugene F. Fama (1960) and Paul A. Samuelson (1960). Remarkably, they independently developed the same basic concept of market efficiency from two rather different research agendas. These differences would drive them along two distinct trajectories leading to several other breakthroughs and milestones, all originating from their point of intersection, the Efficient Market Hypothesis (EMH). The EMH state that in an efficient market where many well-informed and intelligent investors operates, the stock price imitates all the existing information and no other information or analysis can be used to earn abnormal returns.

The arguments of Fama (1965) form the theoretical foundation for the Efficient Market Hypothesis (EMH), which persuasively reasons that in an efficient and active market consisting of many well-informed investors, equity prices will appropriately reflect the effects of information based on present and future expected events. The strong form of the hypothesis asserts that the current market prices fully reflect all private (insider) and public information. In other words, insiders shouldn`t be able to earn excess returns from privileged asymmetric information. The strong form of the hypothesis represents an absolute standard, and in practice, market demonstrates only a certain degree of efficiency.

Efficient Market Hypothesis (EMH) claims that speculative market prices fully and immediately reflect all available relevant information. Fama categorised information as: publicly available information, information that eventually becomes public, insider information. Event studies are used in tests of Efficient Market Hypothesis (EMH) to ask whether prices incorporate information fully on the day that the information is revealed. If Efficient Market Hypothesis (EMH) holds, the information about the event should be incorporated into prices before or on the day of the event itself. There should be no impact on returns after the event

“There was little evidence on the central issues of corporate finance, now we are overwhelmed with results, mostly from event studies”

(Fama, 1991, p. 1600)

Event study analyses are typically used for two different purposes firstly as a test of semi-strong form market efficiency; and secondly as, assuming that the market efficiency hypothesis holds, as a tool for examining the impact of some event on the wealth of firms' shareholders. Event studies measure security price changes in response to events. A single event study typically analyzes the average security price reaction to instances of the same type of event experienced by many firms. For example, the event could be the announcement of a merger. The event date can vary from one security to another in the same study, with dates measured in "event time". Event studies have been used in a large variety of studies, including [mergers and acquisitions], earnings announcements, debt or equity issues, corporate reorganizations, investment decisions and corporate social responsibility MacKinlay (1997), McWilliams & Siegel (1997).

Empirical Evidence

The debate about efficient markets has resulted in hundreds and thousands of empirical studies attempting to determine whether specific markets are in fact "efficient" and if so to what degree. Many novice investors are surprised to learn that a tremendous amount of evidence supports the Efficient Market Hypothesis (EMH).

Since the late 1960s, the enormous study in the finance and accounting literature has recognized evidence of relationship between accounting reports and market reactions. Fama (1970) described an efficient market as having prices that “fully reflect” all available information. Beaver (1981) offers a definition of market efficiency based on the information distribution when investors have mixed beliefs.

Accounting reports probably are one of the sources of public information. Ball & Brown (1968) examine the relationship between the accounting reports & stock prices &. Their results show that the market reacts to unexpected earnings as though the market participants had access to the good or bad news prior to the availability of this news to the market. They estimate that only 10 to15 percent of the market reaction takes place during the announcement month. Using another approach, similar results are also found in the work of Ball and Brown (1968) they examined price changes surrounding the announcement of a firm's annual earnings and found that the stock market reacts quickly to annual earnings announcements. Ball (1992) and Bernard & Thomas (1989) and (1990), documented significant delays in the adjustment of stock prices to quarterly earnings announcements.

Developed countries of the world such as the USA, the UK, and Australia, etc. the amounts of researches on Efficient Market Hypothesis are extensive. Fama, Fisher, Jensen and Roll (1969) conducted the first study on semi-strong form of Efficient Market Hypothesis (EMH). They examined the behaviour of abnormal returns at the announcements of stock splits and found that the market reaction is significant prior to the stock split announcement. Jordan (1973) assessed the behaviour of security prices surrounding the quarterly earnings announcements and found that stock market is efficient in the semi-strong form.

In Asia until now some researches has been done. Kong, S. and Taghavi, M. (2006) study the Effect of Annual Earnings Announcements by using daily data on the Chinese Stock Markets they found that a higher than expected earnings announcement leads to a rise in stock returns on days before the news announcement and a fall afterwards another study carried out by Su, D. (2003) he examines the stock price reactions to changes in earnings per share (EPS) in the Chinese stock markets and found that domestic A-share investors do not correctly anticipate the changes in earnings and fail to adjust new earnings information quickly, but international B-share investors can predict earnings changes better than A-share investors. As a result, abnormal returns (ARs) can be obtained by trading on the earnings information, but for A-shares only. An explanation is that most A-share holders are individuals with short-term investment horizon while most B-share holders are large institutions that trade on more detailed and accurate financial information not immediately available to A-share holders. Odabasi, A. (1998) examine the security returns' reactions to earnings announcements on Istanbul Stock Exchange Using event study on equally weighted portfolio the empirical results indicates that the mean squared excess return (without respect to sign of security returns) on the announcement day is significantly larger than the average during the non-event period. The full sample is also divided into “good” and “bad” news subsamples. The results reveal that the average abnormal returns on announcements days are significantly different than zero for each subsample. These findings are consistent with the prediction that earnings announcements possess informational value. However, the behaviour of the cumulative average abnormal returns do not give full support to the hypotheses that the security prices come to new equilibrium levels after the announcement of earnings.

In Pakistan so far very limited work has been done on stock market. Karachi stock exchange (KSE) has also shown a fast growth in recent years. KSE-100 Index which moved around 10,000 to 12,000 points in the past decades now crossed 14,000 Mark and still expected to grow further. To the extent of my knowledge in Pakistan the study on earning announcement has not been conducted so far nevertheless some event studies have been conducted including Political Events Affecting the Pakistan Stock Exchanges; An Analysis of the Past and Forecasting the Future by Clark, Masood and Tunaru (2005) and some other including the effect of nuclear test etc.

Karachi Stock Market is an emerging stock market and it is very important to conduct some research about the information efficiency aspect of the market. Market efficiency increases the reliability and credibility of the stock market business. This study is an attempt to fill that gap of stock market research.

In this study, we examine the effect of annual earnings announcement on Karachi Stock Exchange – 100 Index. We conduct our study on annual earnings of the companies as this is compulsory for the listed companies to announce their annual earnings.

This paper utilizes an event study methodology to empirically test the affect of annual earnings announcement using daily stock returns for the companies listed in Karachi Stock Exchange – 100 Index; its objective is to the stock market reaction (KSE) to Annual Earnings Announcements. Abnormal returns are defined as the difference between actual and predicted returns surrounding an event. Cumulative average abnormal returns are the sum of average abnormal returns in a given time period. Brown and Warner (1980), Davidson, Dutia and Cheng (1989) Mitchell, Pulvino and Stafford (2002) each utilize a similar event study approach to examine stock market reactions to announcements.

### Sample

We are conducting event study on KSE – 100 Index for three consecutive years; they are 2005, 2006 and 2007. In this study we have selected 66 companies (see Table # 02) for the year 2005, 68 companies (see Table # 03) for the year 2006 and 60 companies (see Table # 04) for the year 2007. Companies being selected in this study are out of 100 companies listed on KSE – 100 index and traded on Karachi Stock Exchange selection process of the companies are discussed in portfolio construction. Further, the companies having any information for the duration of the event window (-30 days to +30days) that influence their stock price are eliminated. A list of deselected companies is presented for year 2005, 2006 and 2006 in appendix III, IV and V respectively.

### Portfolio Construction

In this study one variable - net profit taken as a base for portfolio construction. A total of three portfolios are constructed for each year included for the study on the base of changes in their annual net earnings. Only those companies are selected for the study, which are having more than 20 percent increase or decrease in the current annual net earnings compare to the Corresponding year net earnings in the previous year. The reason for selecting 20% is to capture the effect of the market as lowering the percentage for selection criteria of the companies, might not show any notable reactions of the share holders or investors in either way. The formula for calculating the percentage changes in net earnings is given as:

(Current annual net earnings – Corresponding annual net earnings for the previous year) / (Corresponding annual net earnings for the previous year) *100

After computing the percentage change in the net earnings for 100 companies we divide them into three portfolios. First portfolio consists of companies which are having positive percentage (more than 20 percent) change and called “plus” portfolio. The second portfolio comprises of companies having negative percentage (less than 20 percent) change and called “minus” portfolio. The third one is called “all” portfolio, which includes all the firms selected as sample for the study.

A total of 194 companies are being selected for the study for a period of three year. In 2005 the ‘plus' portfolio consists of 58 companies, ‘minus' portfolio consists of 08 companies and ‘all' portfolio consists of total 66 companies. In 2006 the ‘plus' portfolio consists of 53 companies, minus portfolio consists of 15 companies and ‘all' portfolio consists of total 68 companies. In 2007 the ‘plus' portfolio consists of 32 companies, ‘minus' portfolio consists of 28 companies and ‘all' portfolio consists of total 60 companies.

### Data Description And Data Sources

The data required to perform this study includes the dates of annual earnings announcements for the 100 companies. That is the dates on which the sample companies announce the financial results of the company. The announcement data is available on the download section of the KSE website http://www.kse.com.pk/, annual earnings (net profits) figures of the 100 companies are also available in same link.

The next set of data includes the stock prices for the event window with this daily index prices also required for the Karachi Stock Exchange – 100 Index for the period of 61 days of event window. This set of data is collected from Reuters and from yahoo finance respectively.

### Issues Concerning Daily Data

Brown and Warner (1985) outlined several issues that should be taken into consideration when choosing an event study model. These will be discussed very briefly, followed by a brief description of excess return measures that may be used in an event study to calculate abnormal returns. Based on the findings of Brown and Warner (1985) an appropriate “Excess Return Measure” will be adopted. Brown and Warner (1980) originally used monthly data, and described several techniques to calculate abnormal activity using an event study approach. Their subsequent work involved the use of daily data, where they describe problems pertaining to daily data. Brown and Warner (1985) note that daily data may exhibit stock returns that are not normally distributed, and this raises the possibility of daily returns exhibiting serial dependence.

However, they conclude that methodologies based on the market model are “well specified under a variety of conditions,” including the use of daily price data (Brown and Warner, 1985). Several other authors (Panayides and Gong, 2002; Davidson, Dutia and Cheng, 1989) have verified that the market model is well specified and provides the most accurate measure of abnormal performance.

### Measuring The Daily Returns

For calculating the returns, both for individual securities and for market we have used the daily closing value on index with the assumption of trading done at the closing value. We used the natural log on the index value, so formula for calculating the daily return used for this study is given as;

Following is the formula:

Rt = Ln [Pt / Pt-1]

Where:

Rt = return on day‘t'

Pt = stock price on day‘t'

Pt-1 =stock price on day‘t-1'

Ln= Natural Log.

Period To Study

The data used in this study were retrieved from the website of Karachi Stock Exchange. This database contains daily observations on stock prices and returns as well as annual financial statement data. Each company had different kinds of earnings announcements. They were mainly interim and annual announcements, but there were a few quarterly announcements too. For the purpose of consistency and comparability, the study focuses only on the annual earnings announcements for the years 2005, 2006 and 2007.

### Econometric Package Used

All the calculation done in this study is by using Excel.

Methodology

To test for the existence of abnormal returns, a benchmark for normal returns in required. An estimation of parameters like alpha and beta is obtained by regressing the index's returns to the stock's returns. The Beta value is the slope coefficient, and is also a measure of the stock's volatility

After the estimation of parameters computations for abnormal returns around the event days are being done, a 61 days event window is selected for the study. The window begins 30 days prior to the event date and ends 30 days after the event. The remaining one day is the announcement date in case where the announcement day is an holiday or the stock market is closed on that day than the following next day is taken as announcement day. Additionally, care has been taken to ensure that, during the parameter estimation period, no other corporate event, such as stock split, mergers and acquisition etc is taking place which may cause abnormal returns.

In the first step, a regression is being carried out using the returns on a given stock and the returns of a stock market index. Assuming a constant Beta value for a given stock, we calculate the estimated return of stocks in the event window period as follows.

E (Rit) = αi +βi Rmt

Where E (Rit) is the expected return at time t, αi and βi are parameters of the regression equation and Rmt is the daily return on a stock market index (Karachi Stock Exchange for this study). The abnormal return is defined as the difference between the actual return on a stock Rit and its expected return, E (Rit). Therefore, the abnormal return of a stock at time t is given as;

ARit = Rit - E (Rit) for i = 1…N

Where,

Rit = Actual Returns

After calculating abnormal returns for each security we add them together across the event days for a total of 61 days and then divide the sum with the number of companies in the sample we will get Average abnormal return the formula for calculating the Average Abnormal Returns (AARs) is given below as;

N

Σ ARit

AAR it = i=1

N

Where,

i = the number of securities in the study;

N = total number of securities.

t = the days surrounding the event-day

After calculating average abnormal returns we will than calculate cumulative of average abnormal returns we will get this by cumulating all the average abnormal returns the formula for CAARs is;

K

CAARt = Σ AARit Where t = -30,...0,...+30.

t = -30

Parametric Significance Test

Parametric significance test is done on cumulative average abnormal returns to find out the average price behaviour of stocks during the event window. AARs and CAARs should be close to zero, if markets are efficient. Parametric ‘t'- test is used to assess the significance of AARs and CAARs. The 5% level of significance with n -2 degree of freedom was used to test the null hypothesis of number of significant abnormal returns after the event day. ‘t' values of AARs and CAARs for the event window will draw our conclusion. The formula for the computation of test statistics of Average Abnormal Return for each day during the event window is;

t = AAR

σ (AAR)

Where,

AAR = Average abnormal return

σ (AAR) = Standard error of AAR

The formula to calculate‘t' statistics for CAARs for each day during the event window is;

t = CAAR

σ (CAAR)

Where,

CAAR = Cumulative average abnormal return

σ (CAAR) = Standard error of cumulative average abnormal return

Formula for computing the standard error is given as:

S E = σ .

√n

Where,

S.E. = Standard Error

σ = Standard Deviation

n = Number of Observations

9.2 NON-PARAMETRIC SIGNIFICANCE TEST

Runs test and the Sign test are being used for this study for non – parametric test, these test are used to avoid the restricted assumption of a particular distribution, which a parametric ‘t'- test makes.

9.2.1 Runs Test

We carried out runs test on ARRs and on CAARS with the help of the following formula;

Where,

μr = Mean number of runs

n1 = Number of positive AARs

n2 = Number of negative AARs

r = Number of runs (actual sequence of counts)

The standard error (S.E.) for the expected no's of runs can be computed by using following formula:

A standardized variable ‘Z' as under can express the difference between actual and expected number of the runs:

Z = r – μr

σr

9.2.2 Sign Test

We carried out sign test on ARRs and on CAARS with the help of the following formula; first we have to calculate the standard error of the proportion (σp):

Where,

σp = Standard error of the proportion

P = Expected proportion of positive AARs = 0.5

q = Expected proportion of negative AARs = 0.5

n = Number of AARs

To compute the value of sign test we used the following equation:

Where,

P = Actual proportion of AARs in the respective years having positive signs

PHo = Hypothesized proportion = 0.5

We calculated sign test statistics before and after the event day (that is during the event window).

### Results / Findings

We have presented our results / findings separately that is year by year first we look at the year 2007 and obtain important findings from its calculations; we also include some essential findings in the form of tables and graph. We replicate the same procedure for rest of the two year that is year 2006 and 2005.

After obtaining results / findings we moved to the next section which is analysis / interpretation of research results this part is also done in the same way that is individually (year by year). Only the conclusion is done together which is done after the analysis / interpretation of research results.

### Results / Findings For The Year 2007

The findings of the study are presented in Figure # 02, Table # 05, Table # 06, & in Table # 07.

The results in Table # 05 exhibits that for plus portfolio during the event window of 61 days, 20 days have negative and 41 days have positive AARs under market model with log returns. This means that about 32.79% of the days have negative AARs and 67.21% of the days have positive AARs. In order to have more clarity, CAARs are presented in Table # 05 and in Figures # 02. CAARs are positive for 96.72% of the event days and negative for just 3.28% of the event days. This suggests that CAARs are positive for 59 days out of 61. This result indicates that the market expected good news in the annual earnings announcement and the same is conveyed in the earnings.

For minus portfolio, AARs are positive for 56.67% and 53.33% of days before and after the event days respectively. Of the 61 days, AARs are negative for 28 days and positive for just 33 days. This means that AARs are negative for 46.67% of days and positive for remaining 53.34% of days. This implies that for minus portfolio; AARs are almost identical.

The results in Table # 05 and Figure # 02 revealed that CAARs are negative for 13 days and positive for 17 days before the event day. After the event day CAARs are negative for 11 days and positive for 19 days. Of the 61 days, it is negative 24 days (39.34%) and positive for 37 days (60.66%). A close look at Table # 05 reveals unexpected results. Both AARs and CAARs are positive for the majority of the days before and after the event day for minus portfolio.

On the other hand for ‘all' portfolio, out of 61 days, AARs are negative for 26 days and positive for 35 days. This shows that AARs are negative for more than 42.62% of days and for the remaining days are negative.

In the case of CAARs, the results presented in Table # 05 and in Figures # 02 indicate that CAARs are positive for 58 days and provide an opportunity to earn abnormal returns on the basis of annual earnings announcements.

Notes:

1. ‘Plus' Portfolio: The firms with positive 20 percentage or more change in net earnings i.e. earnings after tax.

2. ‘Minus' Portfolio: The firms with negative 20 percentage change or more in net earnings i.e. earnings after tax.

3. ‘All' Portfolio: All the firms selected as samples for the study.

4. AARs show the values of average abnormal returns.

5. CAARs show the cumulative average abnormal returns, which are computed for days -30 through 30.

6. Day -30 to -1: The days before the annual earnings announcement.

7. Day 0: The day of the annual earnings announcement.

8. Day 1 to 30: The days after the annual earnings announcement.

### Runs Test

‘When the expected number of run is significantly different from the observed number of runs, the test rejects the null hypothesis that the daily returns are random. The run test converts the total number of runs into a Z statistic. For large samples the Z statistics gives the probability of difference between the actual and expected number of runs. The Z value is greater than or equal to +1.96, reject the null hypothesis at 5% level of significance' Sharma and Kennedy (1977).

As can be seen from the table # 06, the Z statistics of AARs is less than +1.96 and positive, this mean that the runs values are not significant at 5% level during the event windows. Therefore, we accept null hypothesis that AARs occur randomly.

Note:

The average abnormal return is statistically non-zero at the 5% significance level. Where runs and sign test statistics are greater than the critical value of ±1.96.

Test

### Results / Findings For The Year 2006

The results in Table # 08 exhibits that for plus portfolio during the event window of 61 days, 25 days have negative and 36 days have positive AARs under market model with log returns. This means that about 40.98% of the days have negative AARs and 59.02% of the days have positive AARs. In order to have more clarity, CAARs are presented in Table # 08 and in Figures # 03. CAARs are positive for 86.89% of the event days and negative for 13.11% of the event days. This suggests that CAARs are positive for 53 days out of 61. This result indicates that the market expected good news in the annual earnings announcement and the same is conveyed in the earnings.

For minus portfolio, AARs are positive for 53.33% and 46.67% of days before and after the event day respectively. Of the 61 days, AARs are negative for 31 days and positive for 30 days. This means that AARs are negative for 52.83% of days and positive for remaining 49.18% of days. This implies that for minus portfolio AARs are nearly equal throughout the event window.

The results in Table # 08 and Figure # 03 revealed that CAARs are negative for only 01 day and positive for 29 days before the event day. After the event day CAARs are negative for again 01 day and positive for 29 days. Of the 61 days, it is negative for just 02 days (3.28%) and positive for 59 days (96.72%).

In the case of overall portfolio, out of 61 days, AARs are negative for 26 days and positive for 35 days. This shows that AARs are negative for more than 42.62% of days and for the remaining days are positive.

In the case of CAARs, the results presented in Table # 08 and in Figure # 03 indicate that CAARs are positive for 58 days and provide an opportunity to earn abnormal returns on the basis of annual earnings announcements.

Notes:

- ‘Plus' Portfolio: The firms with positive 20 percentage or more change in net earnings i.e. earnings after tax.
- ‘Minus' Portfolio: The firms with negative 20 percentage or more change in net earnings i.e. earnings after tax.
- ‘All' Portfolio: All the firms selected as samples for the study.
- AARs show the values of average abnormal returns.
- CAARs show the cumulative average abnormal returns, which are computed for days -30 through 30.
- Day -30 to -1: The days before the annual earnings announcement.
- Day 0: The day of the annual earnings announcement.
- Day 1 to 30: The days after the annual earnings announcement.

### Runs Test

‘When the expected number of run is significantly different from the observed number of runs, the test rejects the null hypothesis that the daily returns are random. The run test converts the total number of runs into a Z statistic. For large samples the Z statistics gives the probability of difference between the actual and expected number of runs. The Z value is greater than or equal to +1.96, reject the null hypothesis at 5% level of significance' Sharma and Kennedy (1977).

As can be seen from the table # 09, the Z statistics of AARs for plus portfolio is greater than +1.96 and positive, this mean that the runs values are significant at 5% level during the event windows. Therefore, we reject null hypothesis for plus portfolio that AARs occur randomly. For minus and all portfolio the Z statistics is less than +1.96 and positive, this mean that the runs values are not significant at 5% level during the event windows. Therefore, we accept null hypothesis for minus and ‘all' portfolio that AARs occur randomly.

### Sign Test

From table # 09 the sign test statistics shows that for ‘plus' portfolio, ‘minus' portfolio and ‘all' portfolio; the computed values are not significant at 5% level. Thus, the sign test for all three portfolio we accepts the null hypothesis that there is significant difference between the number of positive and negative AARs.

Note:

The average abnormal return is statistically non-zero at the 5% significance level. Where runs and sign test statistics are greater than the critical value of ±1.96.

Test

### Results / Findings For The Year 2005

For minus portfolio, AARs are positive for 46.67% and 53.33% of days before and after the event days respectively. Of the 61 days, AARs are negative for 30 days and positive for 31 days. This means that AARs are negative for 49.18% of days and positive for remaining 50.82% of the event days. This implies that for minus portfolio AARs are nearly equal during the event window.

In the case of ‘all' portfolio, out of 61 days, AARs are negative for 22 days and positive for 39 days. This shows that AARs are negative for 36.07% of days and for the remaining 63.93% days AARs are positive.

The results for CAARs, presented in Table # 11 & in Figures # 04 indicate that CAARs are positive for 39 days and provide modest opportunity to earn abnormal returns on the basis of annual earnings announcements.

Notes:

- ‘Plus' Portfolio: The firms with positive 20 percent or more change in net earnings i.e. earnings after tax.
- ‘Minus' Portfolio: The firms with negative 20 percent or more change in net earnings i.e. earnings after tax.
- ‘All' Portfolio: All the firms selected as samples for the study.
- AARs show the values of average abnormal returns.
- CAARs show the cumulative average abnormal returns, which are computed for days -30 through 30.
- Day -30 to -1: The days before the annual earnings announcement.
- Day 0: The day of the annual earnings announcement.
- Day 1 to 30: The days after the annual earnings announcement.

### Runs Test

‘When the expected number of run is significantly different from the observed number of runs, the test rejects the null hypothesis that the daily returns are random. The run test converts the total number of runs into a Z statistic. For large samples the Z statistics gives the probability of difference between the actual and expected number of runs. The Z value is greater than or equal to +1.96, reject the null hypothesis at 5% level of significance' Sharma and Kennedy (1977).

As can be seen from the table # 12, the Z statistics of AARs for plus and ‘all' portfolio is greater than +1.96 and positive, this mean that the runs values are significant at 5% level during the event windows. Therefore, we reject null hypothesis for ‘plus' and ‘all' portfolio that AARs occur randomly. For minus the Z statistics is less than +1.96 and positive, this means that the runs values are not significant at 5% level during the event windows. Therefore, we accept null hypothesis for minus portfolio that AARs occur randomly.

### Sign Test

From table # 12 the sign test statistics shows that for ‘minus' portfolio; the computed values are not significant at 5% level. Thus, the sign test for the ‘minus' portfolio accepts the null hypothesis. On the other hand, in the case of ‘plus' and ‘all' portfolio computed sign test values are significant at 5% level. Therefore, on the basis of sign test statistics we reject the null hypothesis for ‘plus' and ‘all' portfolio that there is no significant difference between the number of positive and negative AARs.

Note:

The average abnormal return is statistically non-zero at the 5% significance level. Where runs and sign test statistics are greater than the critical value of ±1.96.

– Test

### Analysis / Interpretation Of Research Results / Findings

Market reaction to annual earnings announcements is empirically tested in this study by selecting the companies listed in Karachi Stock Exchange`s 100 Index. The periods selected for the study are 2005, 2006 and 2007 and the numbers of companies selected to study for each year is 66, 68 and 60 companies respectively.

As we said in the previous section; we have analyse each year separately. Preliminary year is 2007 than 2006 and finally year 2005.

### Analysis / Interpretation Of Research Results / Findings For The Year 2007

The event study methodology of residual analysis is used. The results of the study for the year 2007 reveal that for plus portfolio during the event window of 61 days, AARs are positive for the majority of the days and CAARs are positive for 59 days under market model with log returns. In the case of minus portfolio, AARs and CAARs are positive for the majority of the days. For ‘all' portfolio, AARs are positive for the majority of the days and CAARs are positive for 58 days.

For all the three portfolios computed runs values for AARs are not significant at 5% level during the event window. Therefore, we accept the null hypothesis that AARs occur randomly. The sign test statistics shows that for ‘minus' portfolio and ‘all' portfolio, the computed values are not significant at 5% level. Thus, on the basis of sign test statistics for the ‘minus' portfolio and ‘all' portfolios we accept the null hypothesis that there is no significant difference between the number of positive and negative AARs. On the other hand, in the case of ‘plus' portfolio the computed sign test values are significant at 5% level. Therefore, on the basis of sign test statistics we reject the null hypothesis that there is no significant difference between the number of positive and negative AARs.

The t-test values on AARs for all three portfolios shows that they are significant at 5% level for more than 60% of days during the event window and remaining days they are not significant. This indicates the existence of abnormal returns on daily basis for the majority of the days. The t-values on CAARs are significant for more than 80% of days for all the three portfolios. Therefore, on the basis of t test statistics on CAARs we reject the null hypothesis that CAARs are close to zero.

### Analysis / Interpretation Of Research Results / Findings For The Year 2006

The event study methodology of residual analysis is used. The results of the study for the year 2006 reveal that for plus portfolio during the event window of 61 days, AARs are positive for the majority of the days and CAARs are positive for 53 days under market model with log returns. In the case of minus portfolio AARs are negative for more than 50% and CAARs are positive for 59 days out of 61. For ‘all' portfolio, AARs are positive for the majority of the days and CAARs are positive for 58 days.

For minus and all portfolios computed runs values for AARs are not significant at 5% level during the event window. Therefore, we accept the null hypothesis that AARs occur randomly. For plus portfolio the computed runs values for AARs are significant at 5% level during the event window. Therefore, we reject the null hypothesis that AARs occur randomly for plus portfolio. The sign test statistics for all three portfolios, the computed values are not significant at 5% level. Thus, the sign test statistics for all three portfolios we accept the null hypothesis that there is no significant difference between the number of positive and negative AARs.

The t-test values on AARs for all three portfolios shows that they are significant at 5% level for more than 60% of days during the event window and remaining days they are not significant. This indicates the existence of abnormal returns on daily basis for the majority of the days. The t-values on CAARs are significant for more than 69% of days for all the three portfolios. Therefore, on the basis of ttest statistics on CAARs we reject the null hypothesis that CAARs are close to zero.

### Analysis / Interpretation Of Research Results / Findings For The Year 2005

The event study methodology of residual analysis is used. The results of the study for the year 2005 reveal that for plus portfolio during the event window of 61 days, AARs are positive for the majority of the days and CAARs are positive for 42 days under market model with log returns. In the case of minus portfolio, AARs are positive for just above 50% and CAARs are positive for just 07 days rest of the days are negative. For ‘all' portfolio, AARs are positive for the majority of the days and CAARs are positive for 39 days.

For ‘minus' portfolios computed runs value for AARs are not significant at 5% level during the event window. Therefore, we accept the null hypothesis that AARs occur randomly. For ‘plus and ‘all' portfolio computed runs value for AARs are significant at 5% level during the event window. Therefore, we reject the null hypothesis that AARs occur randomly. The sign test statistics shows that for ‘minus' portfolio the computed value are not significant at 5% level. Thus, the sign test statistics for the ‘minus' portfolio we accept the null hypothesis that there is significant difference between the number of positive and negative AARs. On the other hand, in the case of ‘plus' portfolio and ‘all' portfolio the computed sign test values are significant at 5% level. Therefore, on the basis of sign test statistics we reject the null hypothesis that there is no significant difference between the number of positive and neg

### Cite This Dissertation

To export a reference to this article please select a referencing stye below: