School of Education, Culture and Communication
Division of Applied Mathematics
MASTER THESIS IN MATHEMATICS /APPLIED MATHEMATICS
Market Efficiency Analysis at the Stockholm Stock
Exchange: Measuring Intraday Stock Price Performance
around Interim Reports of the OMXS30 Large Cap
Stocks - An Event Study Approach
By
Juan Marcelo Tames Blanco and Samuel Osei Nsiah
Magisterarbete
i matematik/tillämpad matematik
Division of Applied Mathematics
School of Education, Culture and Communication Mälardalen University
School of Education, Culture and Communication
Division of Applied Mathematics
Master Thesis in Mathematics / Applied Mathematics
Date:
3
rdJune 2010
Project name:
Market Efficiency Analysis at the Stockholm Stock Exchange: Measuring
Intraday Stock Price Performance around Interim Reports of the OMXS30 Large
Cap Stocks - An Event Study Approach
Author(s):
Juan Marcelo Tames Blanco and Samuel Osei Nsiah
Supervisor(s):
Lars Pettersson
Examiner:
Anatoliy Malyarenko
Comprising:
30 ECTS credits
Abstract
The purpose of this thesis is to perform event studies that determine the level of efficiency of twenty-two large cap stocks from the OMXS30 Index. Under the event study methodology, analysts’ expectations, the standard Capital Asset Pricing Model and a set of parametric tests
are implemented. As a result, significant evidence is found on the existence of intraday abnormal returns at the exact moment of an interim report publication. However, further
Acknowledgements
I thank God for this thesis. To all my family especially my mother Elena and my uncle Roberto whose support and belief made me the person I’ve become. Last but not least to the love of my life Lindsay who has become my driving force and has filled my life with happiness. Special thanks to my closest friends Diego R., Franklin, Juan Pablo and Abraham.
- Marcelo Tames
First and foremost I offer my sincerest gratitude to God for His numerous blessings, guidance and protection in the course of this thesis and my education as a whole. I owe my deepest appreciation to Ms. Agnes Owusu my mother for her support and encouragement. I also dedicate this thesis to my late father who I know will be proud of his son wherever his soul maybe. Lastly, I offer my regards and blessings to all of those who supported me in any respect especially Goergina Akoto Nsiah, Esther Nsiah and Gisele Nana Frema Nsiah during the course of my studies.
- Samuel Osei Nsiah
We would also like to thank Lars Pettersson for the guidance and support he provided during this thesis.
Special thanks to Per Högberg from NASDAQ OMX who provided us with the valuable tick-by-tick stock data for the OMX Large Cap stocks from the Nordic Region.
Contents
1. INTRODUCTION ... 1
1.1 Problem Area. ... 1
1.2 Purpose of the Study ... 2
1.3 Limitations of the Study ... 3
1.4 Notation and Terminology ... 4
2. THEORETICAL BACKGROUND ... 6
2.1 The Efficient Market Hypothesis ... 6
2.2 Levels of Efficiency ... 6
2.2.1 Weak-form efficiency ... 6
2.2.2 Semi-strong form efficiency ... 7
2.2.3 Strong-form efficiency ... 7
2.3 Types of Efficiency ... 8
2.3.1 Informational Efficiency... 8
2.3.2 Market Rationality ... 8
2.3 Implications of the Efficient Market Hypothesis. ... 9
2.3.1 Insider trading ... 9
2.3.2 Investors... 9
2.3.3 Companies ... 10
2.4 Event Studies ... 10
2.4.1 Price reaction to new information ... 10
2.4.2 Methodology for an Event Study... 11
2.5 Different approaches to perform event studies ... 14
2.5.1 Information on Analysts’ Forecasts ... 14
2.5.2 Capital Asset Pricing Model – CAPM ... 14
2.5.3 Parametric Tests ... 15
2.6 Previous Results of Some Event Studies ... 18
3. DATA AND METHODOLOGY ... 19
3.1 The data collection and approach ... 19
3.1.1 Analysts Expectations Data ... 19
3.1.2 Exact Report Publication Data ... 20
3.1.3 Tick-to-Tick Stock Price Data ... 20
3.2 Research Methodology ... 22
4. ANALYSIS ... 24
4.1 Analysts’ Expectations and Real Outcomes ... 24
4.1.2 Analysis and Interpretation ... 26
4.2 Capital Asset Pricing Model ... 39
4.2.1 Data Organization ... 39
4.2.2 Analysis and Interpretation ... 41
4.3 Parametric Tests ... 43
4.3.1 Data Organization ... 43
4.3.2 Cross-sectional independence Case ... 46
4.3.3 Constant Standardized abnormal returns Case ... 48
4.3.4 Cross-sectional dependence Case ... 50
4.3.5. Analysis and Comparison of Results ... 51
4.3 Supplementary Information on Stock returns around the event ... 52
5. CONCLUSIONS ... 55 6. REFERENCES
7. APPENDIX
Appendix 1 – Sample SME - Direkt estimation
Appendix 2 – Snapshot of raw data in the filtering process
Appendix 3 – Company-Specific Information based on Analysts’ Expectations Appendix 4 - Company Quarterly Sample Returns
Appendix 5 – Instant Abnormal Returns Appendix 6 – Sample Abnormal Returns
Appendix 7 – Adjusted Company Weights Revised Every Six Months Appendix 8 – Abnormal And Cumulative Returns For Individual Companies
Appendix 9 - Confidence Rejection Probabilities for Average Abnormal Returns (Cross-Sectional Independence case)
Appendix 10 – Confidence Rejection Probabilities for Standarized Abnormal Returns Appendix 11 – Confidence Rejection Probabilities for Average Abnormal Returns (Cross-Sectional Dependence case)
List of Tables
Table 1 - List of the chosen stocks for the analysis. ... 19
Table 2 - Number of Average Abnormal Returns obtained under the three different methods. ... 22
Table 3 - Alfa Laval Analyst’s Estimates against report outcomes. ... 25
Table 4 - Range criteria to represent the level of consensus in rating bars. ... 25
Table 5 - Alfa Laval information on the report release and expected market reaction. ... 26
Table 6 - Range criteria to represent the level of stock price bias ... 26
Table 7 - Descriptive Statistics of Analysts' Sales Deviations. ... 29
Table 8 - Descriptive Statistics of Analysts' Pretax Profit Deviations. ... 30
Table 9 - Descriptive Statistics of Quarterly Returns. ... 32
Table 10 - Company Quarterly Returns. ... 33
Table 11 - Descriptive Statistics of Quarterly Abnormal Returns. ... 34
Table 12 - Range criteria to represent the Efficiency Real Level. ... 35
Table 13 - Efficiency Measure on Sample Abnormal Returns Size. ... 35
Table 14 - Company Quarterly Abnormal Returns. ... 36
Table 15 - Average Quarterly Abnormal Returns. ... 38
Table 16 - Average Quarterly SSVX 1M Interest Rate. ... 40
Table 17 - CAPM-based Average Quarterly Abnormal Return Calculation. ... 41
Table 18 - Expected Quarterly Report Return. ... 42
Table 19 - CAPM and Analysts’ Expectations Model Average Quarterly Sample Abnormal Returns. ... 43
Table 20 - Descriptive Statistics of Average Quarterly 5-Min Returns. ... 44
Table 21 - Average Quarterly Abnormal Returns Statistics – Cross-Sectional Independence case. ... 46
Table 22 - Critical Values for a Two-tailed T-test. ... 47
Table 23 - Average 5-minute Confidence Levels to Reject Ho – Cross-Sectional Independence case. ... 48
Table 24 - Critical Values for a Two-tailed Z-test. ... 49
Table 25 - Average 5-minute Confidence Levels to Reject Ho - Standardized abnormal returns case. ... 49
Table 26 - Average Quarterly Abnormal Returns Statistics – Cross-Sectional Dependence case. ... 50
Table 27 - Average 5-minute Confidence Levels to Reject Ho – Cross-Sectional Dependence case. ... 51
Table 28 - Average Mean and Volatily Behavior Around Report Publication. ... 53
Table 29 - Average Volume and Spread Behavior at different tick levels after report Publication. ... 54
List of Figures
Figure 1 - Efficiency levels based on information. ... 8
Figure 2 - Possible price reactions to new negative information. ... 11
Figure 3 - Average abnormal returns around the announcement date. ... 13
Figure 4 - Average Deviation between Report Results & Analysts Expectations(Mean Pretax Profit Based). ... 27
Figure 5 - Average Deviation between Report Results & Analysts Expectations(Median Pretax Profit Based). ... 27
Figure 6 – Deviation among Analysts’ Estimates as the numbers of number of analysts increases. ... 28
Figure 7 - Relationship between Deviation among Analysts’ Estimates and Deviations in actual results. ... 29
Figure 8 - Frequency Distribution of Analysts' Sales Deviation. ... 30
Figure 9 - Analysts' Pretax Profit Deviation Frequency Distribution. ... 31
Figure 10 - Quarterly Return Frequency Distribution. ... 32
Figure 11 - Quarterly Abnormal Return Frequency Distribution. ... 34
Figure 12 - Quarterly Report-Based "Instant" Abnormal Returns for the OMXS30 Index. .... 37
Figure 13 - Quarterly Report-Based "Sample" Abnormal Returns for the OMXS30 Index. ... 38
Figure 14 - Scatter Plot of Expected Quarterly Report Returns under the CAPM and Analyst’s Expectation Models. ... 42
Figure 15 - Average Quarterly 5-Min Return Frequency Distribution. ... 44
Figure 16 - All-Stock Quarterly 5-Minute Interval Abnormal Returns around Announcement Time. ... 45
Figure 17 - All-Stock Quarterly 5-Minute Interval Cumulative Abnormal Returns around Announcement Time. ... 46
Figure 18 - Average 5-minute Cumulative Rejection Frequency & Confidence Rejection Probability for the three cases reviewed. ... 52
Figure 19 - Quarterly Report-Based Returns for the Studied Sample Size in the OMXS30 Index. ... 53
1
1. INTRODUCTION
Over the years there has been different ways of how investors look at the market to take some specific action in search for profits. Theories have changed and modern financial theory has become one of the most complex theories ever developed. Based on more mathematical and normative models, capital markets are more systematic and have as cornerstone the market efficiency hypothesis. As a result, a market can be grouped into three different levels of efficiency: strong, semi-strong, and weak. Stock prices in a perfectly efficient market reflect all available information. These differing levels, however, suggest that the responsiveness of stock prices to relevant information may vary1. Thus, investors have a wide range of opinions on what the market level of efficiency actually is.
The methodology of event studies provides evidence on the semi-strong form of the Efficient Market Hypothesis (EMH). For that reason, this methodology tests whether publicly available information is fully reflected in current stock prices. When a company releases its interim report, investors are updated on the firm’s performance and future prospects. Based on the new information, they are able to assess and adjust their expectations according to the actual results. Therefore, such event should have an impact in the company´s stock price in the case the results don’t match market expectations and, at the same time, open room for abnormal returns that provide evidence about the level of efficiency the stock price reaction has.
1.1 Problem Area.
Some studies about the efficiency level of the Swedish stock market have been made in the past, but the authors identified the need to make an intraday study which hasn’t been found in the literature review. Having most of the studies on market efficiency performed in markets in the United States, there has been a little evidence from smaller markets like Sweden which may provide a different case. An important study was performed by Forsgårdh and Hertzen (1975) who found that the Swedish Stock market was less efficient than other foreign counterparts. Their study was based on the average adaptation of share prices to new information presented in the company financial statements during a four year period. Lately, Ekdahl and Aram (2003) tested the Swedish stock market for the weak form efficiency and found enough evidence to support the hypothesis. However, Gyllefjord and Lolic (2006) performed a semi-strong test where it was proved that the Swedish market might have some persistent trends following an either positive, or negative interim report publication. Furthermore, during the last years the Swedish market was affected by a global financial meltdown which motivates its research. For that reason, the authors believe also that the time frame selection is particularly important and the choice of using tick-by-tick stock price information would prove to be more accurate when supporting the informational efficiency theory which deals with the speed information is incorporated in the stock price.
If stock markets were strongly efficient, and stocks were always valued correctly, professional investors would find themselves unemployed in the absence of arbitrage opportunities. However, as we can see in reality, this extreme case is out of question and researchers instead try to find models and empirical evidence that could go closer to such ideal market.
1
Investopedia, a Forbes digital company, Retrieved April 15th, 2010, from http://www.investopedia.com/terms/m/marketefficiency.asp
2 Nevertheless, the efficient market hypothesis is controversial and there are several arguments both for and against it. A generation ago, the EMH was generally accepted after Eugene Fama’s leading work in 1970. By then, it was commonly believed that capital markets were exceptionally efficient when reflecting new information into the securities prices and thus, the market itself. However, there has been evidence against market efficiency questioning investors’ rationality and the way there are informed. Even though the theory is supported by random price changes, these variations are proved to be larger than the theory predicts2
.
The semi-strong form of market efficiency suggests that only information that is not publicly available can benefit investors seeking to earn abnormal returns on investments3. In this scenario, the way of how this information is spread and interpreted plays an important role when generating market expectations. Professional Analysts base their work on this information and, given certain assumptions, they fairly represent as a whole what the market expectations are. Furthermore, deviation in those estimates after new information comes into the market can still mean efficiency given the market corrections are in line to what previously was expected. Malkiel (2003) explained that, while we have seen cases when market pricing was not just right at all in cases like the market crash in 1987, the IT bubble a decade ago and lately the housing bubble leading to a global financial crisis, efficiency can still be achieved if investors don’t find a systematic way of making the most of such anomalies to earn above-average abnormal returns without taking above-average risks.
The problem also lies on whether Analysts’ expectations are a good fundamental basis to represent market expectations and, consequently, calculate abnormal returns. For that reason, a parametric test and a more theoretical model such as the standard Capital Asset Model (CAPM) might support or contradict the former results and, as a result, a more general conclusion can be reached in terms of finding evidence about excess abnormal returns around a company interim report announcement.
1.2 Purpose of the Study
The purpose of this thesis is to perform event studies that determine the level of efficiency of twenty-two large cap stocks from the OMXS30 Index. The chosen stocks have been consistently part of the index during a period of 43 months beginning in January 2006 to July 2009. Furthermore, the objective is to prove the existence of abnormal returns between two hours prior and two hours after the publication time of the companies’ interim reports. To do this, analysts’ expectations are used as a measure to verify the intraday speed of stock price adjustments when new information comes into the market. Then, to test the validity of the results, two other methods are used to reach a more general conclusion. First, the standard Capital Asset Pricing Model is used to calculate expected quarterly returns instead of using the analysts’ expectations. Consequently, a comparison is made between the average quarterly abnormal returns in both cases. Finally, a set of parametric tests are run to check the significance level of abnormal returns.
2
Mauboussin J. Michael, 2002. Revisiting Market Efficiency: The Stock Market as a Complex Adaptive System 3
Investopedia, a Forbes digital company, Retrieved April 15th, 2010, from http://www.investopedia.com/terms/s/semistrongform.asp
3
1.3 Limitations of the Study
Before going into the theoretical background and due to the nature of this research, the authors call for the following limitations:
Market Efficiency Dependence: In order to perform the event study methodology it is essential to assume markets are efficient (at least in its weak form). However, this assumption is not valid in many situations. Every single investor might respond to event signals in a different manner and might need different lengths of time to assess new information. Therefore, markets could exhibit market inefficiencies because prices do not instantly or fully reflect all available information.
Short-run impact Study: Although this is beneficial when analyzing the effect of a particular event. The choice of the event window is assumed to give enough time to investors to act upon the company announcement. However, short-window event studies may not accurately capture the whole economic impact of the event and the initial response may be biased, incorrect or incomplete. Even though such error is rectified later, it can be argued that the chosen event-window was not large enough. Nevertheless, this study limits to these boundaries to make inferences and draw conclusions from them.
Incomplete subset of data: Even though the collected data comes from the real source that generates it (NASDAQ OMX), some files that contained subsets of data couldn’t be accessed due to corrupted or missing information. Despite this fact, a total of 306 out of 330 possible interim reports were examined (around 7% of missing information). The first quarter of 2008 and the second of 2009 contained most of the missing data. However, the authors consider that in general, the available data is reliable and provide enough evidence to make inferences and draw conclusions.
Sample Size and Adjusted OMXS30 Index: The selected sample size comprehends twenty-two large cap stocks which make up for what the authors call “Adjusted OMXS30 index”. The reason to adjust the index is because every six months the index is reviewed and some stocks don’t fulfill the requirements to remain the index4. As a result, they are replaced by other stocks. In order to study a consistent market index, the authors sort out the regular stocks that remain in the list during the study period and, at the same time, the weight contribution from the irregular stocks is equally distributed among them. This new index is then considered to represent the market index in a better way.
Irregular Periodicity of publications and correlation between firms: All firms don´t publish their reports on the same day and each firm doesn’t do it exactly every three months. For example, in a certain quarter one firm might be the first presenting its results and the following quarter it might be among the last ones. Furthermore, this fact might influence the behavior of abnormal returns provided that stock price reactions might be correlated among firms that publish their report on the same day and, uncorrelated when the event day is not common to the firms. Therefore, the authors attempt to address this issue by testing the significance of abnormal returns under the conditions of both, dependent (correlated) and independent (uncorrelated) abnormal returns. However, this fact of reality must still be considered when interpreting the results this study reaches.
4
These stocks normally have the smallest market capitals and therefore their contribution to the index is minimal.
4 Business Cycle Effect: The stock market can be seen as a reflection of the economic situation of a country and therefore behaves according to the economic cycle. As a result, the market sentiment and the stock price reactions might be more optimistic in good times and, more pessimistic when the economy goes into a recession. The chosen study period (from 2006 to mid-2009) is intended to reflect a complete business cycle represented by an economic boom in 2006 through the half of 2007 and then, a severe economic recession (financial crisis) from that point towards the end of the study period. Overall, the possible overreactions due to this factor are not considered when implementing the event study methodology. However, it is considered when explaining the possible reasons of stock price anomalies and interpreting the results.
Other Missing Factors: There are other factors that have been proved to historically have an influence in the stock price behavior. These factors are for instance the January effect, weather effect, day-of-the-week effect, momentum effect, etc. Even though they could have facilitated a better understanding of abnormal returns, they are not considered in this study. Sensitive Results: The results of this event study are sensitive to changes in the research design. A different choice of model calculating expected returns can result in significant different results for abnormal returns. The authors approach this issue by estimating expected returns with the constant mean return model and the CAPM apart from analysts’ expectations to reinforce the study and have a more complete view of the stock price behavior.
In spite of all the limitations, the event study methodology has a powerful and easy design which enables to detect abnormal performance. Therefore it can be used in less than perfect conditions and the results are straightforward to interpret and share.
1.4 Notation and Terminology
OMX Stockholm 30 (OMXS30): The Nordic region’s most well known and widely used index, the OMXS30 index, consists of the 30 most-traded stocks on Stockholm Stock Exchange. The limited number of constituents guarantees that all the underlying shares of the index have excellent liquidity, which results in an index that is highly suitable as underlying for derivatives products5.
ISIN code: International Securities Identifying Number
Tick-to-tick data: Tick-by-tick data as used in this thesis refers to any market data which shows the price and volume of every print. It also includes information about every change in the biding and asking prices of traders.6
Stock Liquidity: Explains the level to which a security can be traded without significantly affecting price. A liquid security can experience a high volume of trading without a considerable change in price.7
5
OMX DERIVATIVES MARKETS, Secondary name to Stockholm Stock Exchange Ltd. SE-105 78 Stockholm. SWEDEN, 2007, Efficient Securities Transactions, Retrieved August 23th, 2009 From www.omxgroup.com
6
TRADE IDEAS, a data trading website. Retrieved April 15th, 2010, from http://www.trade-ideas.com/Glossary/Tick_Data.html
7
5 Pretax earnings: Net income before income taxes is deducted.
Coefficient of determination (R2): Statistical measure of how well a regression line
approximates real data points. It is worth also noting that R2 is a nonnegative quantity. The limits of R2 are given as . Where a R2 value of 1 means a perfect correlation and a value of 0 means no correlation.8
Hypothesis testing: An appropriate criteria to find out whether the estimated values or
sample evidence agree with the expectations of the theory being tested. A theory or hypothesis that is not verifiable by appeal to empirical evidence may not be acceptable as part of scientific theory.9
Z & T-tests: Test of significance which is a procedure by which sample results are used to
verify the accuracy or inaccuracy of a null hypothesis10.
Two-tailed test: applied to test the significance procedure in a case where the two extreme
tails of the relevant probability distribution as well as the rejection regions are considered and the null hypothesis is rejected if the computed t-statistics lie in either tail11.
The Central Limit Theorem: The CLT says that the distribution of sample means will be approximately normally distributed regardless of the population distribution.
Event window: The time window where the event takes place.
Kurtosis: A measure of the peakness or flatness of the data in relation to the normal distribution. A high kurtosis implies heavy tails in distribution12.
Skewness: Measures the symmetry of a distribution with respect to its mean. Negative skewness means there is a substantial probability of a big negative return. Positive skewness means that there is a greater than normal probability of a big positive return13.
8
Gujarati, “Basic Econometrics” Fourth Edition, 2004, pp.84 9
See Milton Friedman, “The Methodology of Positive Economics,” Essays in Positive Economics, University of Chicago Press, Chicago, 1953.
10
Gujarati, “Basic Econometrics” Fourth Edition, 2004, pp. 129 11
Gujarati, “Basic Econometrics” Fourth Edition, 2004, pp.128 12
Gujarati, “Basic Econometrics” Fourth Edition, 2004, pp.886 13
6
2. THEORETICAL BACKGROUND
2.1 The Efficient Market Hypothesis
As defined by Eugene Fama 40 years ago, the efficient financial market is one where all available information is fully reflected in the security prices. In that vein, the Efficient Market Hypothesis has been a significant proposition in finance. All the time spent on analyzing, trading and picking securities would all be a waste as the EMH indicates that investors cannot consistently beat the market. If this is true, the market is perfectly priced, and it is better to passively hold the market portfolio and skip active money management14.
The EMH gained a quick support and several amount of studies indicated its influence. Nonetheless, the last twenty years academic researchers have challenged the EMH. Recent studies have actually found evidence against the EMH. This is mostly regarding the factors making the markets efficient, being weaker and more incomplete than previous research has supposed. An important alternative view of the EMH is Behavioral Finance which states that there are systematic and significant deviations from efficiency persisting for longer periods15. The EMH is based on the following three assumptions:
• Presence of rational investors who value securities rationally.
• Trades are random making the existence of irrational traders irrelevant which doesn’t affect security prices.
• Presence of rational arbitrageurs will eliminate the irrational investors influence on prices16.
Some of the evidence against the EMH is that investors are not fully rational. Shleifer (2000) states that investors base their investment decisions on unnecesary information. Arbitrage can be risky as mentioned in behavioral finance since securities do not always have perfect substitutes, therefore arbitrage is limited. Various empirical studies have defied the EMH, showing that volatility in stock market prices are higher than expected and that stock prices overreact which is in line with psychological theory. Evidence has also been found on stock price predictability. There are also many studies finding statistical evidence of under reaction and overreaction in security returns following news and earnings announcements. The implication of the EHM is that if new information is published about a firm it will be incorporated into the share price rapidly and rationally, with respect to the direction of the share price movement and the size of that movement17.
2.2 Levels of Efficiency
Historically, when Fama introduced the concept of the EMH in 1970, the levels of market efficiency are categorized into three forms namely: weak-form efficiency, semi-strong form efficiency and the strong-form efficiency.
2.2.1 Weak-form efficiency
14
Shleifer, Andrei, 2000, Inefficient Markets: An Introduction to Behavioral Finance. 15
Shleifer, Andrei, 2000, Inefficient Markets: An Introduction to Behavioral Finance. 16
Shleifer, Andrei, 2000, Inefficient Markets: An Introduction to Behavioral Finance. 17
7 The EMH in its weak form states that current share prices fully reflect all information contained in past price movements. This makes it worthless to base trading rules solely on share price history, due to the fact that the future cannot be predicted. To test the weak-form efficiency, technical analysts use a wide range of trading rules. To name a few examples, some analysts recommend buying shares that have performed well in relation to the rest of the market, retaining that their performance will continue in that direction. Other analysts also propose a purchase when the price of a share increases at the same time as an increase in trading volume occurs. In any case academicians propose that the weak form of the EMH is to be accepted since the share price history cannot be used to predict the future in any peculiarly profitable manner.
2.2.2 Semi-strong form efficiency
This level of efficiency proposes that all the relevant publicly available information is fully reflected in share prices. This includes not only past price movements, earnings and dividend announcements, rights issues, technological breakthroughs, resignations of directors, and so on18. The implication of semi-strong form of efficiency is that there is no advantage in analyzing publicly available information after it has been published, since the market has already captivated it into the price. Tests of the semi-strong form efficiency place emphasis on the issue of whether it is useful acquiring especial expensive information or analyzing publicly available information. If semi-strong efficiency is true it undermines the work of several analysts whose trading rules cannot be applied to produce abnormal returns i.e. semi-strong efficiency implies already weak form efficiency.
2.2.3 Strong-form efficiency
In this case, all relevant information, public and private, is reflected in the share price. Insider trading is the main focus in this level of efficiency. Few privileged individuals such as major shareholders or the board of directors commonly have access to detailed business information compared to the regular investor in the market. Even though the market is seen as being inefficient at this level, insiders find it difficult to make abnormal profits. However, tests like the one of Asbell & Bacon (2010) have proved that to make abnormal profits, one way is to trade shares on the basis of insider trading announcements. Many people have kicked against insider trading arguing that those outside of the charmed circle feel cheated. The idea here is that some investors feel that profits are being made by these insider traders at their expense. When this happens, most investors are usually not willing to buy shares which have a negative effect on society. To repose confidence in investors, several stock exchanges have punitive measures for insider trading deals if it comes to their notice.
Later on Fama (1991) refers to these levels of efficiency as tests for return predictability, event studies and tests of private information instead of the traditional weak, semi-strong and strong formats respectively. In any case the authors use interchangeably the terms for future references. To sum up, Figure 1 illustrates the three levels of efficiency as a function of information degrees.
18
8
Figure 1 - Efficiency levels based on information.
Figure 1 - Efficiency levels based on information
2.3 Types of Efficiency
Having described the degrees of efficiency a market can have, another way to look at market efficiency is through the concepts of informational efficiency and market rationality.
2.3.1 Informational Efficiency
Informational efficiency refers to the speed in which new information is incorporated into market prices19. This concept of informational efficiency has a wonderfully counterintuitive and seemingly contradictory flavor to it: the more efficient the market, the more random the sequence of price changes generated by such a market must be. This, of course, is not an accident of nature but is the direct outcome of many active participants attempting to profit from their information20. In this situation the random walk model plays an important role. Developed by the French mathematician Louis Bachelier (1900), the random walk model states that price movements will not follow any patterns or trends and that past price movements cannot be used to predict future price movements. Although the stock market may not be a mathematically perfect random walk, the statistical dependencies giving rise to a momentum of arbitrage are extremely small and are not likely to permit investors to realize excess returns. Therefore, empirical evidence supporting the random walk model contribute to evidence claiming that markets are more informational efficient than many would have thought.
2.3.2 Market Rationality
While a great majority focuses on finding informational efficiency evidence, a number of authors emphasize on prices accurately reflecting investors’ expectations based on the present value of future cash flows. This hypothesis is known as market rationality. If markets exhibit rationality, there should be no systematic differences between share prices and the value of the security based on the present value of the cash flow to security holders21. In the case of a
19
Elton, Gruber, Brown, Goetzman, 2007, Modern Portfolio Theory and Investment Analysis, pp. 429 20
Farmer, D. and A. Lo, 1999, “Frontiers of Finance: Evolution and Efficient Markets”. 21
Elton, Gruber, Brown, Goetzman, 2007, Modern Portfolio Theory and Investment Analysis, pp. 429
Strong – Public & Private available information
Semi-strong – Publicly available informtion
Weak – Historical price
9 stock split which doesn´t economically change the value of the company, prices can be shown to respond to such event and thus, be powerful evidence against market rationality. Furthermore, any persistent anomaly like those so-called “market effects22” proves the market rationality hypothesis and therefore the EMH wrong. However, it can be argued that investors take advantage of such anomalies to make profits. Moreover, it is uncertain that patterns can be created out of those anomalies and even if that is the case, tranding costs would minimize this arbitrage opportunities23. Additionally, when an anomaly is found, Malkiel (1973) explains that traders exploiting an inefficiency cause it to disappear. Therefore, market rationality can still be supported since the EMH is concerned with under what conditions an investor can earn excess returns.
2.3 Implications of the Efficient Market Hypothesis
.The efficient market hypothesis has several implications for investors and companies as well as insider traders who contribute to market inefficiency. These implications are outlined below.
2.3.1 Insider trading
Usually insiders earn excess returns to what is previously expected. Unless they have superior analytical ability, their excess return must be due to illegal exploitation of insider information24. Previous works from Jaffe (1974) and Lorie & Niederhoffer (1968) found a pattern claiming that insiders purchase/sell stocks months before an increase/decrease in its value. Therefore, market regulators try to tighten insider regulations in order to prevent investors from being manipulated by insider actions. In Sweden, Finansinspektionen is the responsible agency that deals with this issue. By checking the way insiders trade with their holdings, this agency makes sure that their actions have no effect on the stock price. When Finansinspektionen has reason to believe that a breach of the Market Abuse Penal Act has taken place, the matter is immediately transferred to the Swedish National Economic Crimes Bureau25.
2.3.2 Investors
Most people believe that public information is not a justification to earn abnormal returns. Malkiel (1973) once famously quoted: “a blindfolded monkey throwing darts at a newspaper’s financial pages could select a portfolio that would do just as well as one carefully selected by experts." In other words, this implies that there is no need for fundamental analysis since it will be a waste of money and as long as efficiency is preserved, the average investor should simply select a suitably diversified-portfolio. At the same time, he would avoid costs related to deep analysis and transactions. There is the need for investors to seek a substantial amount of timely information. Therefore, companies should be motivated by investors’ pressure and market as well as government regulations to provide compatible
22
The size effect, the market/book effect, the January effect, and the day of the week effect, etc. 23
Investor Home, Historical Stock Market Anomalies, Retrieved December 20th, 2009 From http://www.investorhome.com/anomaly.htm
24
Elton, Gruber, Brown, Goetzman, 2007, Modern Portfolio Theory and Investment Analysis, pp. 426 25
10 information26. If stricter and more severe punitive measures are introduced for insider dealers, the perception of a level playing ground could be attained.
2.3.3 Companies
Companies take an important part when the market efficiency comes into play. Since the efficiency level depends on the quality of information, corporate managers have the power of choosing which information will be made public. However, the pressure they have may force them to sacrifice long-term gains to short-term profits. This obsession with short-term results by investors, asset management firms, and corporate managers collectively leads to the unintended consequences of destroying long-term value, decreasing market efficiency, reducing investment returns, and impeding efforts to strengthen corporate governance27. Another important issue concerns the timing of security issues when managers sometimes feel that they should wait to take action with the belief that the current market price is undervalued. If they decide to delay the sale, hoping that the market will rise to a more normal level, the idea of the EMH is undermined because the efficient market claims that the such delays shouldn’t exist.
2.4 Event Studies
An event study investigates whether the occurrence of a certain event creates an opportunity for economic profit and, as a result, testes market efficiency at the semi-strong level by looking for evidence supporting the existence of abnormal returns. The event can either have a positive or negative effect on the value of the security, but the study itself examines how fast the information was incorporated in the share price. Common examples of significant events are mergers or take-overs, earnings announcements and new-share issues. If the market is efficient, then the stock price reaction to any of the previously mentioned events doesn’t create a profit opportunity for investors.
Earnings announcements represent one of the most important events for a company and for investors who seek to analyze whether the company’s performance goes in line with expectations. There are several models that calculate expected return on a security. Since most studies find abnormal returns of several percent at the time of the announcement, any way of measuring expected returns will show about the same results unless announcements are clustered on days of extreme market movements28. As a result, it can be inferred that the choice of any model to calculate expected returns is irrelevant when interpreting results in the occurrence of an event.
2.4.1 Price reaction to new information
New information from a company’s interim report corresponds to an important event which should have some effect in the company’s stock price. Ross, Westerfield and Jaffe (2005) categorize a possible reaction to new information into three different types illustrated in Figure 2 (in the case of bad news) and described as follows:
• An immediate price adaptation which reflects in full the new information and no further changes happen unless new information is released. This is the case of a perfectly efficient markets response.
26
Arnold, Glen. 2008. Corporate financial management, pp. 603. 27
CFA Centre for Financial Market Integrity, 2006, Breaking the Short-term Cycle, pp. 1 28
11 • Under reaction which means that stock price falls short in reflecting the information
and it takes longer time for the price to get to an efficient level.
• Overreaction which is the opposite from the previous case i.e. when prices respond to new information with exaggeration and investors push the stock price to an overbought or oversold level.
Figure 2 - Possible price reactions to new negative information.
Figure 2 - Possible price reactions to new negative information
2.4.2 Methodology for an Event Study
Currently there’s no unique methodology to perform an event study. A standard method can be based on the works of MacKinlay (1997) and Elton, Gruber, Brown, Goetzman (2007) and proceeds as follows:
1. Identify the event of interest and the time-window where the security price of the firms involved in the event will be examined. In this situation, it is regular to set the event window larger than the time of interest to check for the stock price behavior around the event. For measuring market efficiency, the results become more significant when using the smallest feasible intervals with evidence on short-horizon tests representing the cleanest evidence on efficiency29.
2. Determine the necessary criteria each firm must fulfill to be included of a given firm in the study. The criteria may involve restrictions on data availability like the Stockholm Stock Exchange. Also restrictions like stocks from a specific industry sector, index, etc.
3. Determine the precise time of the event ocurrence and designate it as zero. In the case for the earnings announcement event, the time the report is released is set as zero.
29
Fama, Eugene, 1991. "Efficient Capital Markets II” Journal of Finance, pp.1602 Stock
Price
Days around the announcement
-20 -10 0 10 20 Delayed Response to bad news Overraction to bad news Efficient market response to bad news
12 4. Compute the actual return for each interval in the event-window being studied for each firm. Stock returns from one point in time to another can be calculated with the following formula:
(1)
Where:
: Stock return at time t : Stock price at time t
: Stock price at time t-1
5. Choose the model to compute the expected return for each firm over the event window. The models can be categorized into two categories: statistical and economic. As it was explained before, any way of measuring expected returns will show about the same results since the variance of abnormal returns is not reduced by much when selecting more sophisticated model and, as a result, showing a lack of sensitivity. Nevertheless, the four main models are described as follows:
a. Constant Mean Return Model: Basically the model is defined in the following formula: (2) And, ! Where: : Mean Return
: Period t-return for security i
: Time period t disturbance for security I with an expectation of zero and
variance !
Although it is the simplest model, evidence from Brown and Warner (1980, 1985) prove that this model often yields similar results to those which are more sophisticated.
b. Market Model: This model is probably one of the most popular among the rest since it includes the market portfolio return to calculate the security portfolio. The formula is illustrated as:
" #$ % (3)
And,
&
& !
Where:
$: Period t-return for the market portfolio
%: The zero-mean disturbance term
"' #' (): Parameters of the market model
This model, also known as the one-factor market model, can prove to be an improved version of the previous one if the of the market regression model is large enough to reduce the variance of abnormal returns.
13 c. Multifactor Models: A multifactor model accounts for more variables in order to explain the stock price behavior. However, previous studies have shown that the gain is small when using this model rather than the previous two described above.
d. Economic Models: When statistical models can´t be run due to constrains in the available data. Economic models can also be used to calculate expected returns. The two most known models are the Capital Asset Pricing Model (CAPM) and the Arbitrage Pricing Theory (APT). CAPM will be explained in more detail in a latter section. However, this model was popular in the early stages of the methodology until significant deviations were found by Fama and French (1996). On the other hand, multifactor models are encouraged by the APT model which has the advantage of eliminating the biases introduced by the CAPM.
6. Calculate the respective abnormal returns for each of the intervals studied for each firm in the sample. The calculation of abnormal results at this point looks obvious since it is defined as the difference between the actual return and the expected return. To be more precise:
* + , (4)
Where:
* : Period t- Abnormal return for the security i
: Period t-return for security i
,: Expected Period t-return for security i
7. Compute for each interval in the event window the average abnormal return for all the firms in the sample. Since individual firms have certain characteristics that can’t be compared. It is often better to look at the average abnormal returns across firms. In that way, the effect of other events can be minimized. Figure 3 illustrates the behavior of the average abnormal returns around the announcement date.
Figure 3 - Average abnormal returns around the announcement date.
Figure 3 - Average abnormal returns around the announcement date Source: Elton, Gruber, Brown, Goetzman, 2007, Modern Portfolio Theory and Investment Analysis, pp. 421
8. Compute the cumulative abnormal return from the start of the period. There are a number of methods of time series aggregation. One method consists on adding together the average abnormal returns (5) while another one, known as the buy-and-hold method, compounds the average abnormal returns and use it as a performance measure (6). The following formulas illustrated what has been described:
Excess return
Announcement day
14 -* ... / *0 .... 1 (5) And -* ... 23 4 *... 556 + 0 1 7 8 55 (6) Where: t: Time interval 9:
... : Average cumulative abnormal return across firms :
.... : Average abnormal return across firms
Despite of the difference among methods, it is useful to analyze the behavior of abnormal returns for certain subsets in the event window.
9. Analyze and draw conclusions from the results. At this point, the results of the analysis are examined and conclusions are drawn.
2.5 Different approaches to perform event studies
This section will describe the theoretical framework of three different approaches the authors use to test for semi-strong market efficiency: Analysts’ forecasts as market expectations, Capital asset pricing model to calculate expected returns over the event window and, parametric tests to check the significance level of abnormal returns.
2.5.1 Information on Analysts’ Forecasts
Professional analysts’ forecasts can act as a good and simple measure of what the market as a whole expects from a company in terms of performance. There have been numerous works on analysts’ expectations and most of them were conducted analyzing US firms. Many authors examine whether the content of an analyst forecast contains information already incorporated into the stock price. Givoly and Lakonishok (1979) found that financial analysts' forecasts do have information content and later, Fried and Givoly (1982) found that analysts’ forecasts influence market expectations and stock prices. Furthermore, they indicate that prediction errors of analysts are more closely associated with security price movements, suggesting that analysts' forecasts provide a better surrogate for market expectations than forecasts generated by time-series models30. However, studying the content of analysts’ forecasts, there is also the risk of suffering selection and survivorship biases. The selection bias occurs when the access of information of the set of studied forecasts is controlled. The survivorship bias occurs when the selection of the organization to be studied is based on knowledge concerning past information skill31. With that into consideration, Fried and Givoly (1982) found that the performance of analysts’ estimates may be explained by a set of factors such as the analysts’ reliance on information released after the end of the fiscal year and, more significant, the broadness of the information set employed. Overall, it can be said that their forecasts fairly represent what the market expects based on publicly available information.
2.5.2 Capital Asset Pricing Model – CAPM
30
Fried, D. and D. Givoly, 1982, Financial Analysts’ Forecasts of Earnings: A Better Surrogate for Market Expectations, pp. 82-107.
31
15 One way to understand how the capital markets function is through the Capital Asset Pricing Model. The CAPM explains the risk - expected return relationship which is very important when pricing financial securities. In this regard, the model determines the expected return an investor can obtain from an investment. This model states that the expected return of a security or a portfolio equals the rate on a risk-free security plus a risk premium. Thus, it is not worth embarking on an investment if the expected return does not meet or beat the required return. Generally CAPM says that investors need to be compensated in two ways: time value of money and risk32. The model is written in the form as follows:
; < #8 ;$+ < (7)
Where:
=>The risk-free interest rate
?@: Company Beta
.A: Expected Return of the market
The risk-free interest rate is understood as the money value over time. The other half of the formula represents risk and calculates the amount of compensation the investor needs for taking on additional risk33. In any case, CAPM is an equilibrium relationship.
Analyzing abnormal returns, it is important to point out what the CAPM takes as anomalies. High Beta stocks are expected to give a higher return than low beta stocks because they are more risky34. Damodaran (2002) defines anomalies as patterns in the market behavior which have not any rational explanation. According to Fama & French (1996), patterns in average stock returns that are not explained by the Capital Asset Pricing Model (CAPM) are anomalies.
2.5.3 Parametric Tests
In the vein of non-parametric tests, parametric tests provide valuable information when testing for efficiency in capital markets. The parametric tests described in this section are from Serra (2002) who explains models from Brown and Warner (1980, 1985), with and without independent abnormal returns across firms and Patell’s (1976) standardized residual test.
Under the assumption that individual firm returns are normally distributed, systematically nonzero abnormal security returns that persist after a particular type of corporate event are inconsistent with market efficiency35. Therefore the authors define the null and alternative hypothesis as follows:
BC: The mean abnormal return (sometimes referred to as the average residual, AR) across
firms at time t is equal to zero.
BD: The mean abnormal return across firms at time t is different to zero.
32
Investopedia, a Forbes digital company, Retrieved April 15th, 2010, from http://www.investopedia.com/terms/c/capm.asp
33
Investopedia, a Forbes digital company, Retrieved April 15th, 2010, from http://www.investopedia.com/terms/c/capm.asp
34
Elton, Gruber, Brown, Goetzman, 2007, Modern Portfolio Theory and Investment Analysis, pp. 291 35
16 This hypothesis will then allow testing for the existence of abnormal returns at any point within the event window and thus, check the behavior of the average abnormal returns around the announcement time. The standard statistic is derived as follows:
We begin with the traditional t-test statistic,
E FG + HI C JK Where,
FG: Sample mean HC: Population mean
I: Sample Standard deviation K: Sample size
Since the null hypothesis states zero average abnormal returns, then HC , then it is known that the sample mean for abnormal returns at time k is :....L. Finally the term in the denominator is equal to an estimate of the standard deviation of the average abnormal returns M:.... As a result the test statistic for testing the significance of abnormal returns at point k L
is:
O**....N
N
...P
Q:... is then calculated according to three different conditions that are explained below. L
2.5.3.1 Cross-Sectional Independence36
Cross-sectional independence is assumed when the abnormal returns for each firm at time k are not correlated. The estimate of standard deviation of the average abnormal returnsM:... L is calculated as follows: M:... ML R/ :S@1 @L T U TV M S @1 :@L M:... W L TV M S @1 :@L X Q:... L
Where N is the number of firms.
In order to calculate the standard deviation of the average abnormal return for each security (σ(AR i0)) it is necessary to take into consideration the standard deviation of the time series of
abnormal returns of each firm during the event window (T), the formula for a single firm is calculated as follows:
36
17 Q:@ W/ Y: @+/ :Z [@ \ 1 Z + ] ^
Where d is equal to one plus the number of factors in the market model. Given the null hypothesis previously stated, the statistic in (8) is distributed as Student-t with T-d degrees of freedom which depend on how the standard deviation of abnormal returns is estimated.
2.5.3.2 Standardized Abnormal Returns37
Every firm has its own characteristics and for that reason, its stock price reacts differently when an event occurs. When the averages of abnormal returns are calculated, it may be intuitive to standardize the outcomes to ensure that each abnormal return will have the same variance. This is achieved dividing each firm’s abnormal residual by its standard deviation (obtained over the estimation period); each residual has an estimated variance of 1. The standardized abnormal returns are given by:
:_@L Q::@L
@
In a particular event time k, the test statistic of the hypothesis that the average standardized abnormal returns across firms is equal to zero, is computed as:
` a::........L_
L_
4b6/ :_c1 @L
a:....L_
Considering independence across firms and that :@ are identically distributed, :_@ are assumed to be distributed as unit normal and the standard error of the average standardized residuals is given by:
a:....L_
JTd
By the Central Limit Theorem, the statistic in (11) is distributed unit normal for large N.
2.5.3.3 Cross-Sectional Dependence38
When the abnormal returns for each firm at time k are correlated, Brown and Warner (1980) suggest that the standard deviation of average residuals should be estimated directly from the time series of the average abnormal returns over the estimation period.
Q8:....C W/ Y / :S1 @ T +:....8[ \ 1 Z + ] e
Where 37Serra, 2002, Event Study Tests – A brief Survey, pp. 5 38
18 : ....8 / Y/ : @ \ 1 Z [ S 1 T f
Given the null hypothesis previously stated, equation (13) goes directly to the denominator of the test statistic in (8) which is distributed as Student-t with T-d degrees of freedom which depend on how the standard deviation of abnormal returns is estimated.
2.6 Previous Results of Some Event Studies
As it was stated before, event studies search for evidence on abnormal returns around the announcement date. From this examination, it is then possible to look for some patterns which can lead to an investor making a profit out of it. Even the most cursory perusal of event studies done over the past 30 years reveals a striking fact: the basic statistical format of event studies has not changed over time39. Some works like from Kraus and Stoll (1972) examined the market reaction of announcements informing the purchase or sale of securities and, found supporting evidence on efficiency. Firth (1975) found a substantial increase in cumulative excess returns on the day of the announcement. Nevertheless, Firth shows that most of this increase occurs between the last trade before the announcement and the next trade. For that reason, investors with no private information don’t have enough time to profit from the price change. Overall, this evidence is consistent with market efficiency. Furthermore, event studies based on dividend announcement by Pettit (1972) found that the market seems to adjust rapidly to new information. In addition, similar works on dividend announcements support Pettit’s findings. The other finding of interest to investment professionals is that for a number of types of announcement, investigators have found a long-term drift in abnormal return (called post- announcement drift)40. This evidence shows that firms that took over smaller firms have significant abnormal returns on average over the next five years. However, one must be careful when making inferences on long-horizon event studies since even using the best methods the analysis of long-run abnormal returns is treacherous41. In Sweden, some event studies have been performed like the one from Forsgårdh and Hertzen (1975) who studied the average adaptation of share prices to new information presented in the company financial statements during a four-year period. In their work found they found that the Swedish Stock market was less efficient than other foreign counterparts. Additionally, Gyllefjord and Lolic (2006) performed a semi-strong test based on earnings announcements compared to analysts’ expectations using daily data for a period of four years. They found that the Swedish market might have some persistent trends following an either positive or negative interim report publication. Overall, the EMH at a semi-strong level seems to hold rather well over the years but the mere presence of abnormal returns around events calls for more evidence refuting the possibility of abnormal return patterns which lead to arbitrage.
39
Kotari and Warner, 2006, Econometrics of Event Studies, pp.8 40
Elton, Gruber, Brown, Goetzman, 2007, Modern Portfolio Theory and Investment Analysis, pp. 426 41
19
3. DATA AND METHODOLOGY
3.1 The data collection and approach
The data required to perform the intended event study has followed a systematic scheme in order to grant an impartial research. The study period begins in January 1st, 2006 and ends in July 31st, 2009. The OMXS30 index is composed by the 30 most traded stocks on the Stockholm Stock Exchange. The stock index is reviewed twice a year when some stocks are dropped from the index while others take their place. To ensure consistency within the study period of this research, the authors have sorted out the stocks which have not moved from the list during the study period. Table 1 illustrates the list of the chosen stocks.
Table 1 - List of the chosen stocks for the analysis.
Table 1 - List of the chosen stocks for the analysis
The purpose of this thesis is to check the intraday stock price behavior around company quarterly reports. Thus, the authors perform the analysis for fifteen quarters which starts from the fourth quarter of 2005 (which is published by companies in the beginning of 2006) to the second quarter of 2009. The report publication dates vary for the twenty-two stocks analyzed. Therefore, several dates are examined for every quarter in order to check the behavior of the companies’ stock price in their corresponding report dates.
3.1.1 Analysts Expectations Data
In order to measure the stock price effects. Market expectations are represented by professional analysts’ forecasts. For that reason, the authors have used the services from a specialized agency. SME Direkt delivers the best product when it comes to estimates of pretax earnings, sales, earnings per share, P/E ratios on a corporate, sector and national level as well as market expectations on the economy and currency movements42.
Normally, SME Direkt issues their estimation letters at least one week prior the report publication. Due to sector and firm specific characteristics, each company has different businesses which cannot be used as a basis for comparison. However, two important and
42
Nyhetbyrån Direkt, Retrieved February 25th, 2010, from http://www.direkt.se/Direkt/Estimates.aspx
ABB Sandvik
Alfa Laval SCA B
Assa Abloy B SEB A
Atlas Copco B Securitas B
AstraZeneca Sv. Handelsbanken A
Electrolux B Skanska B
Eniro SKF B
Ericsson B Swedish Match
Hennes & Mauritz B Tele2 B
Nordea Bank TeliaSonera
Nokia Oyj Volvo B
20 common measures among all firms are turnover and pretax profit which are selected for this research. Furthermore, when looking at the pretax figures, it is feasible to take the minimum, maximum, median and mean estimates calculated from all the analysts who took place in the prediction. Finally, the number of analysts involved in the prediction is extracted for a later correlation study between number of estimates and accuracy of the forecast. This procedure was performed for each of twenty-two stocks analyzed in each of the fifteen quarters studied. Refer to Appendix 1 for a sample SME - Direkt estimation.
3.1.2 Exact Report Publication Data
To check for the most precise moment when the report is exposed to the public, the authors have used the historical information newsfeed at Avanza Bank webpage used for trading. Avanza provides access to all the latest news and press releases about the market and specific companies. Avanza information is synchronized with the Direkt news agency which is closely related to SME-Direkt. Direkt provides daily market information in the form of newsletters and company press releases for investors. By cooperating with the world’s leading newswires, Direkt also reports and informs about international events and gets its own news distributed throughout the world43. For this research, it is assumed that the instant the report is published; it is also the moment when the Avanza news feedback shows the key figures used as parameters for comparison. The registered specific time is then used as a basis to calculate the investigation time frame. That is two hours before and two hours after the report publication time. Finally, after the report is published, the authors take the headline of the following analysts’ reaction elaborated from the Direkt news agency as a parameter to understand the market reaction.
3.1.3 Tick-to-Tick Stock Price Data
This study is carried out using intraday tick-to-tick data which was obtained from the original source i.e. NASDAQ OMX Stockholm AB. The data agreement confined the authors the use of large cap Swedish stocks information from January 1st, 2006 until July 31st, 2009. The obtained dataset consists of tick-by-tick snap shots of all order books at the Large Cap NASDAQ OMX markets. One observation corresponds to one occurrence in a given order book. That is an order is inserted, deleted or amended (updated), which causes a given change to the quoted depth and/or prices etc. in an order book. Details on the order, which caused the change, plus the new overall status of the order book post, are recorded as an observation. Statistics on the quoted orders are summed up on the first 20 tick levels, plus as residual group for level above that, on either site of the spread44.
The dimensions of the amount of data being handled were extreme but given such task, it was rewarding at last when it was filtered and ready to act upon the intended event study’s methodology. Nowadays, an average trading day in the large cap sector comprises an estimated cash flow of 20 Billion Swedish Crowns45. In 2006, the corresponding figure was lower but it increased as the market grew in size. Overall, the following scheme was elaborated to systematically sort out the required information:
- The files received from OMX were unzipped using WinZip.
- The unzipped files are then split into one-gigabyte files with the software A.F.6 SPLIT
43
Nyhetbyrån Direkt, Retrieved February 25th, 2010, from http://www.direkt.se/Direkt/NewsAgency.aspx 44
NASDAQ OMX, August 2008, Nordic Exchange Tick data – Orderlog Product description. 45
