The Performance of Gross-
Profit to Asset on the Swedish Stock Market
A comparison to Book-to-Market and Earnings-to- Price in a time frame of 1994-2013
Authors:Emde, Larissa Yildirim, Cem
Supervisor:Äijo, Janne Student
Umeå School of Business and Economics Spring Semester 2016
Master thesis, 15hp
This thesis examines the performance of portfolios sorted by gross-profit-to-asset (GPA) as a quality investing on the Swedish stock exchange. It constructs long-only portfolios and long-short portfolios sorted by GPA, book-to-market (B/P) and earnings- per-price (E/P). Thus, the thesis includes quality and value investing. The thesis compares separately the constructed long-only and long-short portfolios among each other. The long-only strategies are additionally compared to the market index. The study further examines a combined portfolio, sorting for GPA and B/P in order to test Novy-Marx’s findings. He reports, that the average return improves, while the standard deviation remains at the same level for a combined portfolio sorting for GPA and B/P.
This requires a negative correlation.
The comparison is based on different portfolio measurements as i.e. s.d. The asset pricing models CAPM and 5-Factor Model are applied. In addition, actual returns, excessive return over the risk-free rate and over the market index as a benchmark are assessed for the portfolio. The analysis is conducted for the time period 1994-2013 and separately for downturns, considering 2000-2003, 2007-2009 and 2010.
The results show a great applicability of the gross-profitability ratio on the Swedish market. This quality strategy convinces not only during normal times with the portfolios GPA-h (long-only) and GPA-hl (long-short) but also in stressed times. GPA-h reports positive (abnormal) returns GPA-h during downturns. The long-only and long-short portfolios based on GPA outperform the market in both time periods. GPA-sorted portfolios perform in general better and the two value strategies during normal times and downturns, based on the annual average return. Examining the two value strategies EP-sorted portfolios are superior over BP-sorted portfolios. EP-portfolios achieve better performance during downturns, regarding Jensen’s alpha. It can be derived, that EP is countercyclical. The combined portfolio generates high return and has a high standard deviation. The assessed statement of Novy-Marx cannot be confirmed for the Swedish stock market. It has to be stated that we detected positive correlation instead of negative correlation.
It can be derived, that GPA ratio is applicable on the Swedish market, considering the assumptions and limitations of this study. EP-based portfolios show a good performance during downturns. BP- based portfolios do not perform well on the sweidish market in the assessed time frame. The combined portfolio GPABP-hh does increase returns with constant standard deviation, referred to BP-h.
Our findings show, that both value strategies do not outperform the market index. The
EP-based value portfolios outperform BP-based portfolios. EP-h performs better during
downturns considering Jensen’s alpha.
II I. TABLE OF CONTENTS
1. INTRODUCTORY CHAPTER ... 1
1.1. PROBLEM BACKGROUND ... 1
1.2. RESEARCH GAP ... 2
1.3. RESEARCH PURPOSE ... 3
1.4. RESEARCH CONTRIBUTION ... 4
1.5. RESEARCH QUESTIONS ... 4
1.6. LIMITATIONS OF THIS STUDY ... 5
1.7. THESIS DISPOSITION ... 6
2. RESEARCH METHODOLOGY ... 8
2.1. CHOICE OF SUBJECT ... 8
2.2. PRECONCEPTIONS ... 8
2.3. RESEARCH PHILOSOPHY ... 9
2.3.1. Ontology ... 9
2.3.2. Epistemology ... 10
2.4. RESEARCH APPROACH ... 11
2.5. METHODS ... 11
2.6. RESEARCH DESIGN ... 13
2.7. DATA SOURCES AND LITERATURE REVIEW ... 13
2.8. RELIABILITY,REPLICATION AND VALIDITY ... 15
2.9. ETHICAL AND SOCIETAL ISSUES ... 16
3. THEORETICAL FRAMEWORK AND LITERATURE REVIEW ... 17
3.1. EFFICIENT MARKETS HYPOTHESIS AND THE RANDOM WALK... 17
3.2. BEHAVIOURAL FINANCE ... 19
3.3. RISK AND RETURN ... 20
3.4. MODERN PORTFOLIO THEORY ... 21
3.5. REFERRED ANOMALIES ... 23
3.6. RELEVANT ASSET PRICING MODELS ... 24
3.6.1. CAPM ... 24
3.6.2. Fama & French’ 5-Factor Model ... 25
3.7. RELEVANT RATIOS ... 26
3.7.1. Earnings-to-Price ... 27
3.7.2. Book-to-Price ... 27
3.7.3. Novy-Marx’ Gross-Profitability Ratio ... 28
3.8. CHOSEN INVESTMENT STRATEGIES ... 29
3.8.1. Value Investing ... 30
3.8.2. Quality Investing ... 31
3.8.3. Quality and Value Investing ... 32
3.9. LITERATURE REVIEW ... 33
3.9.1. Novy-Marx’ Gross Profitability Premium ... 33
3.9.2. Value Premium - Price-to-Book ... 35
3.9.3. Value Premium - Price-to-Earnings ... 36
4. PRACTICAL METHOD ... 37
4.1. CHARACTERISTICS ... 37
4.1.1. Sample Data ... 37
4.1.2. Time Period ... 38
4.1.3. Market Index ... 38
4.2. DATA ANALYSIS &PROCEDURE ... 39
4.2.1. Collection and Generating Data ... 39
4.2.2. Portfolio Construction ... 39
4.2.3. Calculation of Returns and Abnormal Returns ... 41
4.2.4. CAPM – Practical Contribution ... 42
4.2.5. 5-Factor Model – Practical Contribution ... 42
4.2.6. Hypothesis Testing ... 43
4.3. METHOD AND PROCEDURE CRITICISM ... 45
5. EMPIRICAL FINDINGS ... 46
5.1. RETURN ... 46
5.1.1. Normal Period ... 46
5.1.2. Downturns ... 48
5.2. ABNORMAL RETURN... 49
5.2.1. Normal Period ... 49
5.2.2. Downturns ... 50
5.3. FAMA &FRENCH -5-FACTOR MODEL ... 50
5.3.1. Normal Period ... 51
5.3.2. Downturns ... 52
5.4. HYPOTHESIS RESULTS ... 53
6. ANALYSIS ... 56
6.1. ANALYSIS OF RESULTS AND DISCUSSION ... 56
6.1.1. Long-only Portfolios ... 56
6.1.2. Long-short Portfolio ... 57
6.1.3. Multi-ratio Portfolio ... 59
6.2. DISCUSSION OF 5-FACTOR MODEL,CAPM AND THEORIES ... 60
7. SUMMARY AND CONCLUSION ... 61
7.1. ANSWERED QUESTIONS ... 61
7.2. REMARKS ON THE STUDY ... 62
7.3. FURTHER RESEARCH ... 62
7.4. OVERALL SUMMARY ... 63
IV II. APPENDIX
Appendix 1 - Downturn ... XV Appendix 2 - Regression Analyses ... XVI Appendix 3 - Outputs – Long-only Portfolios ... XVIII Appendix 4 - Outputs – Long-short Portfolios ... XXIX Appendix 5 - Outputs – Mixed-ratio portfolios ... XL Appendix 6 - Correlation ... XLIII Appendix 7 - Comparison Long-only Portfolios and Market ... XLIV Appendix 8 - Tables Long-only Portfolios ... XLV Appendix 9 - Tables Long-short Portfolios ... XLIX Appendix 10 -Tables Mixed-ratio Portfolios ...LII Appendix 11 - Tables Market Index ... LVIII
III. LIST OF TABLES
Table 1 - Results per portfolio in normal period. ... 47
Table 2- Results per portfolio during downturns. ... 48
Table 3 - Results 5-Factor Model normal period... 51
Table 4 - Results 5-Factor Model downturns. ... 52
IV. TABLE OF FIGURES Figure 1 - Research Methodology. ... 12
Figure 2 - Upwards Sloping Random Walk. ... 18
Figure 3 - Efficient Frontier….. ... 22
Figure 4 - Cumulative returns. ... 46
Figure 5 - Answered Research Question. ... 62
V. LIST OF EQUATIONS Equation 1- CAPM ... 24
Equation 2 - 3-Factor Model. ... 25
Equation 3 - 5-Factor Model. ... 26
Equation 4 - E/P Ratio. ... 27
Equation 5 - B/P Ratio. ... 28
Equation 6 – GPA. ... 28
Equation 7 – Return. ... 41
Equation 8 - Jensen's Alpha. ... 42
V VI. LIST OF ABBREVIATIONS
CF/P Cash-flow to price
EMH Efficient Market Hypothesis
GPA Gross-profit to asset
MPT Modern Portfolio Theory
s.d. Standard Deviation
1. Introductory chapter
This introductory chapter gives a brief description of the problem background covered in this thesis. The research gap identifies the area other researchers have not covered in their studies. This gap is seen to be complemented by this thesis and is followed by the definition of the research questions as well as the contribution to literature. In addition, this chapter covers the description of the purpose of this research and limitations.
1.1. Problem Background
Stock markets offer different possibilities for earning interest on invested capital. Often, investors are aiming on gaining excessive return accounting for the taken risk (Garvey, 2005, p 94). When investing into the stock markets investors can follow a range of different investment strategies and asset classes, keeping the idea in mind to achieve the highest return for the lowest risk possible (Markowitz, 1952, p. 79). The suitability of a strategy on a stock market can differ depending on the characteristics of the stock market and its included stocks This contains the consideration of ratios as book-to-price (B/P), earnings-to-price (E/P) or gross profit-to-asset (GPA). The B/P ratio relates the book value (from the balance sheet) to the market value of the company. E/P ratio relates the earnings of the company to the market value of the company. GPA shows how much gross profit a company earns per asset unit (see. chapter 3.7).
Applying a strategy on a market often inherits the exploitation of market anomalies1
an analogue theory is the idea of efficient markets (EMT). Fama (1970, p. 414) explains efficient markets stating that correct prices of securities are traded on the market inherit full information as well as the correct value of the asset. Nevertheless, it is often stated that it is hard to reach full information in all prices, due to several reasons. The theory of EMT is explained in more detail in section 3.1 of this very thesis.
The various investment strategies investors can choose from contain i.e. indexing, momentum trading, quality, value, growth investing where only the quality and value are assessed in our research (Bender et al., 2013, pp. 1-3.). Value investing (see also chapter 3.8.1.) is mentioned first by Graham & Dodd in 1934. It describes the identification of undervalued shares by applying a ratio that includes information of an accounting measure in relation to the quoted price. The measurements utilised in this strategy for this research are book-to-price (B/P) and earnings-to-price (E/P) (Lakonishok et al., 1994, pp. 1547-1548). Applying the value strategy means to detect the company of several companies with the highest value related to the price. Thus companies with higher earnings or higher book values are preferred to companies with lower earnings or lower book value quoting the same price-per-share (Sanan et al., 2013, pp.175-176).
Quality investing (see also chapter 3.8.2.), aims at detecting a high-quality company by analysing certain ratios as Return on Invested Capital (ROIC), financial strength or gross profitability of the company. The latter is covered by Novy-Marx´ paper “The
1 Anomaly is the occasion if the real result does not match the expected calculated result applying a theoretical model. The existence of an anomaly goes along with the evidence of the model being insufficient of describing the real world/market characteristics/events (Schwert, 2003, p. 939)
other side of value” published in 2013. Quality investing plays an important role in our analysis as the study of Novy-Marx is taken as the basis for our conducted research.
In his paper Novy-Marx (2013a) defines the gross profitability ratio as gross profit minus cost of goods sold divided by total assets. His strategy contains the investment in companies with a high profitability ratio that reflects high profitability and thus, high quality (Novy-Marx, 2014, p. 8).
Novy-Marx (2013a) conducted a study analysing and comparing the cross-sectional predicting power of gross-profitability, P/E and P/B ratios on U.S. equities’ return. He found that gross-profitability has roughly the same predictive power of cross-sectional returns as book-to-market. Additionally, he detected a negative correlation of value and profitability measurements and thus, combined the two strategies by building a hedged portfolio (Novy-Marx, 2013a, pp. 20-21).
Novy-Marx’ (2013a) findings and further research that was already conducted on this field provide a great foundation for our research which covers i.e. the analysation of value and quality investing on the OMXS Stockholm stock exchange.
1.2. Research Gap
Several researchers already investigated and compared value stock performance. Basu (1975, 1977, and 1983) for instance discovered some evidence for abnormal returns of stocks with high E/P ratios in comparison to stocks with low E/P ratios in the United States.
Nevertheless, Novy-Marx (2013a; 2013b) compares strategies sorted by gross profitability i.e. with B/P and E/P ratios in the United States. Further, he constructed a portfolio consistent of stocks from 19 different countries in order to provide international evidence. Novy-Marx observed a profitability spread on international level that is even higher compared to the results concerning the US markets. Furthermore, he discovers the profitability spread to be higher than the value spread on the same market.
Since then, the so-called gross profitability premium2
has been topic of some studies.
Liu (2015) and Kisser (2014) investigated the drivers and reasons for the gross profitability premium. Liu (2015) argues that behavioural bias of investors causes the gross profitability premium; while Kisser (2014) states that a company’s higher operating leverage is related to higher risk and thus creates the premium.
Bunsaip (2014) and Zaremba (2014) confirm the existence of a gross profitability premium in their sample data. Bunsaip (2014) tests it on the Thai stock exchange and concludes that profitable firms earn significant abnormal returns compared to unprofitable firms. Zaremba (2014) concludes the same for the 11 Central Eastern European countries he has included in his analysis. Sun et al. (2015) tested the gross profitability premium in 41 different countries (including Sweden) and verify the presence of a gross profitability premium in many countries. Anyhow, Sun et al.
detected a substantial variation across countries. These studies demonstrate that the gross profitability premium seems to be in general a global phenomenon.
2 The premium is the amount the actual return differs from the expected return.
Novy-Marx (2014) has captured approx. 48 years (1963-2010) in his sample period, including market downturns as the dotcom bubble started in 1995 or the latest financial crisis in 2007. Novy-Marx did not examine those downturns separately in detail.
However, Lam et al. (2014) or Liu (2015) pay closer attention on downturns in their studies concerning the performance of the Novy-Marx gross profitability premium.
Hereby, they do not investigate the performance during downturns in Sweden, which identifies a gap. Anyhow, in order to fulfil the gap we conduct an additional analysis of the concerned strategies during market downturns in order to verify Novy-Marx findings to hold also during market downturns on the Swedish market.
As former mentioned and referring to our literature research (see chapter 3.9.) studies concerning the gross profitability ratio on the Swedish market are already conducted.
Nevertheless, to the best of our knowledge and belief we found no study examining a long-short3
portfolio based on GPA in comparison to a long-short B/P and E/P ratio on the Swedish market. However, as former explained similar research is realised including stocks from the U.S. stock market. The presence of differences between Swedish and U.S. stock market characteristics (e.g. size) underlines this very gap. Complementing studies of the relatively big U.S. stock market with a research considering the relatively small Swedish stock market might reveal interesting findings.
To conclude, we identified a research gap concerning the analysis of a strategy sorted by the Novy-Marx profitability ratio during downturns as well as the comparison of strategies sorted by the Novy-Marx profitability ratio, B/P- and E/P ratio on the Swedish stock market. This is covered by this research. Additionally, we construct and test a multifactor portfolio (B/P and GPA) on the OMXS Stockholm stock exchange.
The performance of the portfolios during downturns will be analysed separately from non-downturn periods in order to see how the different investment strategies perform in different market situation.
1.3. Research Purpose
The research purpose encompasses the analysis of portfolios constructed by sorting companies listed on the OMXS Stockholm stock exchange by Novy-Marx profitability ratio, B/P- and E/P ratio in order to provide evidence on the performance of these strategies. More detailed, this thesis aims at allocating information of the performance of the several strategy also compared to the OMXS Stockholm stock exchange fort long-only portfolios, regarding the examined time span.
Conducting this comparison also reveals the behaviour of a portfolio constructed by the gross-profitability ratio on the Swedish stock market and thus, a statement of how the gross-profitability ratio works on smaller stock markets (rather small market cap with approx. US$578 billion) NOCH COMPARISON can be derived.
Both, long-only and long-short strategies are compared among each other. Additionally, we will compare the long-only strategies to the Swedish stock market as a benchmark.
Also, it is tested if the finding of Novy-Marx (2013a) is working as well as on the Swedish stock market.
3 Long-short and high minus low (HML) will be treated as the same in the following
By following the main purpose, the reader gains a deeper understanding of value- and quality investing. Hereby, the thesis gives the reader an understanding of the gross profitability measure. Furthermore, financial theories are covered broadly by this thesis.
Additional focus is given to the dotcom-, financial crisis and sovereign debt crisis in the sample data period. Hereby, we follow the purpose of comparing the performance of the above named strategies during market downturns, which delivers useful and practical information for investors.
The thesis encompasses the time span from 1994-2013. Based on the findings of the quantitative research this degree project will show how portfolios based on the chosen investment strategies and price multiples performed on a historical approach and deliver results based on that.
1.4. Research Contribution
The contribution of this thesis can be divided into theoretical contribution and practical contribution. The study complements the findings of Novy-Marx by providing results for the Swedish market and thus, a small stock market. Furthermore, the examined research includes a separate analysis of the behaviour of the strategies in downturns, which has not been explored in Sweden yet. It enriches the knowledge about the Novy- Marx gross profitability ratio on the Swedish market. The contribution inherits also a statement about the level of applicability of Novy-Marx (multi factor portfolio B/P and GPA) on the Swedish market. In addition, results about the performance of long-short strategies are derived.
Investors who hold stocks in the Swedish stock market might benefit from the results of the examined strategies. If Novy-Marx (2013a) findings are true for the Swedish market, value investor could benefit extremely by adding the gross profitability investment strategy on top of their existing value strategy (based on P/B). Furthermore, the downturn-analysis can contribute the management of portfolios during unstable times and thus, minimise portfolio losses.
The theoretical contribution includes an additional broad idea of the suitability of current financial theories as the CAPM, EMT, Behavioural Finance and can be considered to contribute to researches concerning market anomalies.
All in all, this study encompasses the analysis of three different theories by taking into account current financial theories applied on a small market. It can be seen, that not only investors or portfolio managers, but also analysts and financial theorists can benefit from the consultation of the results. From a theoretical point of view, the research can be consulted as it gives further ideas regarding of the applied theories and models.
From a practical point of view, the thesis offers valuable findings about the gross profitability ratio as well as value investing applied on the Swedish market.
1.5. Research Questions
Novy-Marx (2013a) tests the GPA on the US stock market. For example he compares
the GPA to other ratios and also the benefits of a joint portfolio of GPA/BP. We test the
GPA on the Stockholm stock exchange (NASDAQ OMX Stockholm) and compare it to
BP, EP as well as to the market. Further we examine the performance of a joint portfolio (GPA/BP) on the Swedish stock market. Our research questions are tested on the sample period of 1994 to 2013.
Q1. Do portfolios, which are separately constructed by applying the gross profit-to asset ratio, B/P-ratio and E/P-ratio each beat the market?
Q2. Do portfolios, which are separately constructed by applying the gross profit-to- asset ratio, B/P-ratio and E/P-ratio each beat the market during the crisis?
Long-only and Long-short
Q3. Does a portfolio constructed by applying the gross profit-to-asset ratio beat portfolios that are separately constructed considering value strategies based on B/P- and E/P-ratios?
Q4. Does a portfolio constructed by applying the gross profit-to-asset ratio beat portfolios that are separately constructed considering value strategies based on B/P- and E/P-ratios during the crisis?
Multi-ratio and Long-only (BP-h)
Q5. Will the performance of a portfolio based on a value strategy sorted by B/P-ratio increase, by adding a quality strategy based on gross profit-to-asset ratio (both long-only) without adding additional risk measured in standard deviation (s.d.)?
1.6. Limitations of this Study
It is important to state limitations as they might influence the interpretation of the conducted study and blur results (Price & Murnan, 2004, p. 66). This thesis incorporates different kinds of limitations, which can be categorised as theoretical, sampling and data, methodological and tools as well as limitations referring to abilities. It was not possible for us to overcome those limitations due to several reasons as a limited time frame of conducting and writing the thesis or it was just beyond our control, that are further explained.
The theoretical limitations consider the contributing theories we highlight during our thesis. These theories are included in order to have common accepted comparable measurements (e.g. 5-Factor Model, CAPM). It has to be stressed, that the theories themselves are not tested and do only have a complementing function. Nevertheless, the results gained by including the theories might help other researchers who actually test the theories. However, the theories follow certain assumptions (which are explained in the corresponding section). Those assumptions are fixed when applying the theory and should be considered by the interpretation of the results. Further only two value strategies are tested (based on E/P- and B/P-ratio). Therefore an overall statement about the performance of value investing on the Swedish market cannot be derived.
Furthermore, it has to be stated that Novy-Marx discloses additional findings, which are
neither considered nor tested within the scope of this thesis (as e.g. testing the
predicting power of GPA on the Swedish market).
This thesis considers only listed companies of the OMXS Stockholm stock exchange during 1994-2013. We therefore analyse historical data. This should be considered when deriving statements for predicting purposes. Financial firms are blended out due to differences in financial statement interpretations (Fama et al., 1992; Asness et al., 2013). Thus, any interpretation and adoption of results should keep this in mind. When comparing the results with other markets the difference in markets, industries and companies should be considered. In addition, the interpretation of price multiples can differ from industry to industry, but due to the fact, that we want to keep the portfolios diversified, we do not separate industries in our portfolio formation.
Additionally, transaction cost will not be considered, even though they are important for investors. The main reason for this is that different banks and brokers have all different fees that make it difficult to keep an overview of all. Transaction cost are also faded out, because Sharpe (1964, p. 434) and Jensen et al. (1972, p. 1) blend them out when calculating CAPM and Jensen Alpha. Capital gains and dividends are taxed differently from country to country. In order to apply the results to other small and similar markets like Sweden taxes are not considered within this thesis. Because of likely tax-difference, this thesis does not consider tax.
The limitations resulting of the chosen methods and tools consider the handling with different analysing tools. Regression analysis has e.g. some assumptions, which have to be followed during the analysis. During the analyses outliers are erased and the data is log-transformed for some returns. This blurs the real information content of the data.
We used Excel to sort and construct the portfolio. In order to minimise the error source we constructed the tables using cell references. However, it cannot be ruled out that the constructed tables are faultless. The regression analysis is conducted with SPSS. A connected limitation is the ability and knowledge on the side of the authors. It has to be stated, that SPSS was not familiar at the beginning of the thesis. Thus, non-knowing the software slightly hamper the procedure and analysis.
1.7. Thesis Disposition
This sub-chapter gives a short overview about the contents of every chapter, in order to provide a better understanding regarding the structure of this thesis. chapter one is already discussed at this point. For comprehensive reasons we describe it in the following either way.
Chapter 1: Introductory Chapter
The first chapter brings the reader closer to the topic of the thesis. Therefore, it starts with a description of the problem background. After, the research gap and therefrom, resulting research questions are presented. This is followed by explaining the purpose, as well as the theoretical contribution of the thesis. The first chapter closes by stating limitations and providing a thesis disposition supporting a clear overview of the structure and chapter contents on the part of the reader.
Chapter 2: Research Methodology
The second chapter deals with the research methodology of the thesis. At first it
describes the authors’ reasons choosing this subject, followed by the presentation of the
preconceptions. Furthermore, the research philosophy of the thesis is presented from an ontological and epistemological point of view. Further, business approach, methods and research design are illustrated. The chapter continues explaining Data Sources as well as Reliability and Validity issues. The chapter ends with Ethical and societal issues.
Chapter 3: Theoretical Framework and Literature Review
The third chapter pursues the goal of providing the reader with necessary theoretical background and an overview about previous literature. Therefore, the Efficient Market Hypothesis and Random Walk, behavioural finance, Risk and Return relation, the Modern Portfolio Theory, the Capital Asset Pricing Model (CAPM) and the Fama &
French 5-Factor Model are explained. Furthermore, the price multiples P/E and P/B and GPA are described. These are the base of the later described portfolio construction. This is followed by the explanation of Quality and value investing as well as the background of the combined strategy consistent of a particular value and quality strategy. The literature review is divided into three sub-chapters considering Novy-Marx gross profitability premium and value premium, whereas the value premium is divided into price-to-book and price-to-earnings ratio.
Chapter 4: Practical Method
The fourth chapter is sub-divided into characteristics and data analysis. The former introduces the sample data, time period and market index. The second part illustrates practical information regarding the data analysis and procedure. It gives detailed information about data collection, portfolio construction, calculation of returns, the utilisation of asset pricing models, correlation and the assessed hypothesis. This chapter closes with criticism of method and procedure.
Chapter 5: Empirical Findings
The fifth chapter encompasses the presenting of empirical findings. It is divided into four sections. The first three are Return, Abnormal Return and 5-Factor Model. Those chapters are sub-divided into normal period and downturns, reporting the performance of the portfolios. Last but not least the hypotheses are assessed.
Chapter 6: Discussion
Chapter 6 includes simply two discussion chapters. It first analyse the results by also comparing them to previous research, which is presented in chapter 3.9. This chapter assess all groups of portfolios separately. Chapter six closes with illustrating discussion points regarding the 5-Factor Model, CAPM and theories.
Chapter 7: Conclusion
This chapter is the last chapter of this research and concludes the important information
provided by the study. Besides answering the derived research questions it presents
remarks and suggests further research. An overall summary ends this thesis.
2. Research Methodology
This chapter starts by demonstrating reasons, which influenced the authors to conduct this research. It is followed by the authors’ perception. It continues examining a methodology framework, considering philosophies, approach, design and methods. It further illustrates the reliability of the conducted study as well as ethical and societal issues. It has to be stated that each topic discussed in the following sub-chapters inherit different approaches which inherit a deeper discussions within the very fields. We are aware that they exist and that lines of distinction are often blurred. To the scope of this thesis the following sub-chapters will not conduct a deep discussion of the different approaches but take the most common accepted ideas into account.
2.1. Choice of Subject
Saunders (2012, p. 30) describes that a sufficient research topic has to fulfil certain criterions considering areas as e.g. time frame, data access, interest, and consistence with requested requirements, relevant evidence. Considering especially these criterions we came to the conclusion that our chosen topic combining Novy-Marx’ profitability ratio and the Swedish market is appropriate to examine. We have access to the needed data through the Umeå University Library, which is crucial for our study. Further we are positive about the given timeframe, to the best of our knowledge it is possible to conduct the research within the given timeframe. In addition, as far as our assessment outreaches we surmise that it is consistent with the thesis manual of Umeå School of Business and Economics.
The authors’ field of interest is reflected by their specialisation of their master program in Finance. Searching for some research ideas, too many have been insufficient regarding the different criterions mentioned above. The examined research idea evolved consulting different theories; having supporting supervisor help and critically reviewing literature (see also chapter 2.7). Nevertheless, this research question attracts our attention due to its topicality and its contribution to prosperous portfolio constructions.
Further, we perceive the study as a support and interest for our future work-life.
Additionally, Novy-Marx’ established profitability ratio by thinking outside the box and not just to rely on currently accepted ratios left an impressed mark on us. Furthermore, it is interesting to examine how and if different investment strategies can be applied on different markets with various characteristics, detecting new findings. Thus, applying the Novy-Marx ratio embedded in the topic of portfolio strategy applied on the Swedish market would further contribute to current research and thus, has a benefit not only for our curiosity but also for other researchers.
As the choice for a subject can be seen partly as a subjective (by the concerned researchers) implication it is important to consider authors’ specific preconceptions and how they might affect the further process of research (Bryman & Bell 2011, p. 31).
Those implications can have a positive effect, as e.g. the choice of subject is taken
according to the authors’ interest or negative effects as e.g. they may entangle evidence
by miss-interpreting findings (Bryman & Bell 2011, p.31). It can be stated that a
research will never be completely free of preconceptions. Nonetheless, measurements
can be established in order to confine extensive, negative perceptions (Bryman & Bell 2011, p. 30). To give a short example, this sub-paragraph about preconceptions may be totally biased by the reflected personal values combined with the individual understanding of the read chapters by the reader in order to construct this paragraph.
However, to prevent this paper from being affected by personal values measures as co- reading, considering a significant amount of different opinions and applying statistical tools and numbers based on common accepted calculation methods as foundation for our interpretation. Furthermore, quantitative research is based on objective data (see chapter 2.6. and is therefore less prone to subjective personal influences. Thus, as our research follows quantitative methods, personal-value based preconceptions are less affecting findings than e.g. using qualitative methods. It is important to state that the awareness of preconception is taken into account during the whole process of establishing/ interpreting the research itself and writing the thesis.
2.3. Research Philosophy
The methodological position in terms of research philosophy clarifies the author’s approach regarding the transmission of the concerned new findings and the applied presumptions on it. To not extent the scope of this paper we focus on epistemology and ontology. It can be stated that the two types shape and influence the conduction and interpretation of the research depending on which position is adopted (Saunders, 2012, pp. 124). Saunders (2012) further states, that especially in business related research a clear and correct philosophy does not exist. Thus it is important to be aware of this very dissent while following the position. The philosophical view is inevitable for the research process (Saunders, 2012, p.126; Bryman & Bell, 2011, p. 24). The following two sub-chapters explain and derive the tended view on ontology (describes the chosen view on reality) and epistemology (deals with the knowledge itself) considered in this research.
Ontology describes the perceived assumption on reality, which is applied on the concerned study and affects i.e. the used formulation of the research questions, the research process and evidence display (Bryman & Bell, 2011, p. 23; Saunders, 2012, p.
127). Thus, it can be argued that it is important to examine the social entities (here the researchers) relation towards the subject. Bryman & Bell (2011, p. 20) distinguish between objectivism and subjectivism.
Objectivism encompasses externality of the social actor towards the concerned topic (whether by conducting, observing etc.), meaning that the existence of the concerned phenomena or entity is non-dependent on other subjects or entities as the social actor but rather exist without the need of a definition of the social actor (Saunders, 2012, p.
128; Bryman & Bell, 2011, p. 21).
Subjectivism on the other hand perceives reality as a consequence of the concerned
social actors influence (being through e.g. his behaviour or thinking). This ontology
approach is also called constructionism, following the idea that reality is constructed by
the linked social actor. It encompasses nominalism, which embraces the assumption that
studies and researches are a construction of the researcher’s interfering (Saunders 2012,
p. 129). It can be seen that in terms of constructivism, reality is a continuous process depending on the interfering of social actors (Bryman & Bell, 2011, p. 22).
This paper perceives the ontological position of objectivism, perceiving the reality as external existent no matter of the authors’ perceptions or influence. This is expressed by conducting an empirical analysis (see chapter 4.) using quantitative methods in order to gain an unbiased objective view on the examined external reality of the behaviour of i.e.
the Novy-Marx profitability measure on the Swedish stock market. This philosophical position also refers to our understanding of preconceptions regarding the examined research and evidence. The data itself is gathered externally and hence, is not influenced by the researchers. An evolving subjective influence could be detected by sampling stocks for the portfolios. This selection is underpinned with other researches, in order to highlight a non-subjective influence. Further an additional bias may arise by interpreting the findings. It can be argued that preconceptions are prevented by stating our findings beforehand in chapter 5.
Epistemology affects the whole process of research as it reflects the treatment of knowledge (Saunders 2012, p. 127). It considers areas as validity, acceptance, communication and legitimation of the perceived and used knowledge (Burrell and Morgan, 1979, cited in Saunders 2012, p. 127). Bryman & Bell (2011, p. 15) describe epistemology as the accurate treatment of generated or included knowledge within a certain study area.
Positivism refers to the application of natural science practices in social science and inherits both elements of inductive and deductive approach (see chapter 2.4). Bryman &
Bell subdivide positivism into the following assumptions which have to be fulfilled:
knowledge are confirmed phenomena (by senses; principle of phenomenalism), hypotheses have to be set up and explainable; (the principle of deductivism), knowledge is derived from facts as the basis for evolving laws (the principle of inductivism), objective conduction, differentiation between scientific and normative statements assuming the former to be “true” (normative could not be confirmed by senses) (Bryman & Bell, 2011, p. 15). Critics concerning this definition of positivism lie in the assumption of deriving methods from the natural science for social research (Bryman &
Bell, 2011, p. 16). To not extend the scope of this paper and due to a broad acceptation by other researcher the former explained idea of positivism is taken into consideration.
Further distinctions are Realism and Interpretivism which are briefly demonstrated. The former shares some attributes of positivism. It encompasses despite the objectivism view the imitation of natural science research procedures (Bryman & Bell, 2011, p. 16).
The approach of realism can be divided into empirical and critical realism (Bryman &
Bell, 2011, p. 17). Interpretivism on the other hand, reflects the idea that the researcher's view is important and influential for the study process. Hence, it distinguishes social science from natural science using the idea that people’s inference on the conducted research is crucial. It therefore requires distinctive methods from natural science taking into account people's behaviour and intuition (Bryman & Bell, 2011, pp. 16).
As the conducted research within this paper does not seek to interpret human behaviour
but rather find and explain phenomena with methods of natural science with an
objective treatment through number based (statistical) support the positivism approach
is utilised. This can be underlined by the application of the former explained assumptions of positivism. The predefined hypotheses (see chapter 4.2.6) are constructed and explainable by the results (deductivism); facts are derived from objective data. Further the objectivism is given by the ontological position derived in the previous chapter. The epistemological position of positivism can be derived from several characteristics of the ontological position objectivism, as i.e. the objective connection from the actor towards research based phenomena is clearly linked to the objective application of analysing data.
2.4. Research Approach
The research can be studied with an inductive or deductive approach (Bryman & Bell, 2011, p. 15). Deductive, in terms of methodology is concerned with the testing of a certain theory empirically (Bryman & Bell, 2011, p. 11), while the inductive approach examines the relation of actors and subjects (Zikmund et al., 2010, p. 633).
The deductive view constructs a hypothesis as the researcher takes a given theory in order to test it (Saunders, 2012, pp. 144). Hereby the difficulty is the deduction of a hypothesis “and then translate it into operational terms”, as Bryman & Bell (2011, p.
11) point out. Thus, the combination of gathering data and hypothesis is important, as well as the data collection itself, and the testing of the hypothesis. In order to conduct the research with sufficient findings the position towards the theory has to be indisputable and defined (Saunders, 2012, p. 51). Zikmund et al. (2012) describe the deductive view as “the logical process of deriving a conclusion about a specific instance based on a known general premise or something known to be true” (Zikmund et al.
2012, p. 651), whereas the inductive approach connects the findings to the investigation of the relation and certain evidence (Zikmund et al. 2012, p. 633).
The inductive approach still incorporates theories impacting the field of study, in order to generate theories from observed relations (Saunders 2012, p. 51). Thus, it can be concluded, that the deductive view takes generality in form of a theory and test it towards a specific hypothesis, whereas the inductive approach generates theory by observing relations.
Our study follows the deductive approach as we are transferring theory concerning i.e.
the profitability ratio of Novy-Marx partly on the Swedish stock exchange. This is done by constructing precise hypothesis, which are further empirically analysed. In addition, as our study does not generate a new theory but test the Novy-Marx profitability ratio compared to “traditional” price- multiples it can be stated, that it follows the deductive approach.
The used approach for methods can be derived from the epistemology position and can be divided into a qualitative and quantitative approach. The latter refers to the natural science (positivistic) epistemological and to the external reality (ontology) approach.
The former, by contrast, follows the interpretivism epistemological approach and
perceives the subjective, ontological view (Bryman 1984, pp. 76; Bryman & Bell 2011,
p. 27). According to Bryman & Bell, quantitative research embraces methods of natural
science to construct social science research and therefore follows an objective reality.
Quantitative research conducts the analysing and gathering of the data in a quantitative manner following a deductive approach. Thus, the quantitative approach can be used in order to verify given theories (Bryman & Bell, 2011, p. 27), by assessing them with fixed, empirical tools (Zikmund et al., 2012, p. 134; Bryman, 1984, p. 78).
Contrary, the qualitative strategy gathers information rather with words than with quantification of external data, following the steady adjusted reality based on the interfering of social actors. Thus, this approach follows the positivism as explained in chapter 2.2 and inherit the inductive approach. Thus, this approach is favourable in order to derive new theories (Bryman & Bell, 2011, p. 27).
The examined study follows the quantitative approach in order to test the derived hypothesis. Thus, this study is related to theory and former research (see chapter 3) under account of different circumstances that have not been tested before (see chapter 1.2., like the testing of GPA on the Swedish stock market. We conduct the study by the usage of several measurements and testing hypothesis in order to pursue an objective outcome as it is also supported by previous studies following a similar approach (see chapter 3.11.). The research methods are the practices the researcher uses in order to gather the relevant data (Bryman & Bell, 2011, p.41). For a quantitative, deductive methodology an empirical, statistical test is more suitable to applied (Bryman & Bell, 2011, p. 41). The practiced methods for this research encompass empirical tests following the quantitative approach. Thomson Reuters is taken as the tool to gather the relevant, non-subjective data.
Figure 1 illustrates the main approaches considered for this research regarding research methodology, marked in red.
Figure 1 - Research Methodology.
2.6. Research Design
The research design is important for the process of the study as it gives a frame for the continuing process. It includes the relation between the chosen variables, the adopted philosophy and the reliability, replication and validity of the research (Bryman & Bell 2011, p. 40). The latter can be seen as attributes of the research design and will be discussed in the next sub-chapter. However, Bryman & Bell distinguish between five different research designs. The comparative design compares two or more distinctive circumstances using the same methods in order to underpin the statement and can be of quantitative or qualitative nature (Bryman & Bell 2011, p.63). Case study design is the comprehensive analysis of a certain subject (Bryman & Bell 2011, p. 59). The longitudinal design refers to changes in the examined data in a process view and derives a conclusion of the findings within a time frame (Saunders 2012, pp. 200). This design is rarely introduced as it is considered to be costly and time consuming (Bryman & Bell 2011, p. 57). An experimental design is the experimental evidence generation and has a solid internal validity (Bryman & Bell 2011, p. 45). Last but not least the cross-sectional design is also referred to social survey. It includes methods as structured observations, official statistics, diaries and content analysis. It is described as the data collection on more than one situation and usually compared to a certain point of time in order to quantify evidence comparing to more than one variable using the quantitative approach (Bryman & Bell 2011, pp. 53; Saunders 2012, p. 200) design which has to be considered regarding the study investigation (Bryman & Bell 2011, p. 40).
Our research is mainly based on the cross-sectional research strategy. This is due to the cross-sectional nature of the return observation, as the different phenomena regarding the profitability ratio of Novy-Marx are measured at a particular time. It can be argued, that time is hence, a delimitation of the study as the evidence is only valid at the examined date referred to the comparable time frame. However, the study is also partly longitudinal as it examines the time of the crisis separately and compares the results to the “normal” times. As the focus of the study is primary on the cross-sectional comparison we consider our research mainly cross-sectional.
2.7. Data Sources and Literature Review
Literature .review is important in order to get familiar with the background of the specific area to research, showing your capability to understand and critically reflect theory and previous studies (Bryman & Bell, 2011, p. 91). The review is seen to help the author to structure and underpin ideas with findings from previous studies and critically assess given examples establishing a profound argumentation (Saunders, 2012, p. 70). The literature review can be based on scientific papers, books and magazines concerning topics of interest for the research area (Zikmund et al., 2012, p. 654).
Reviewing literature can be conducted by different approaches, where an important one is the systematic review. This approach is seen to be fundamental for studies who are aiming at testing to generate particular evidence. The usage of this approach increased lately in business (Bryman & Bell, 2011, p. 94). This sort of review aims at giving the researcher a guide and basis for his own research. Moreover, it is seen to be scientific. A contrary review is the traditional narrative review (Bryman & Bell, 2011, pp. 94-96).
Saunders (2012) on the other hand, distinguishes between three approaches. Preliminary
search is conducted in order to find first ideas and underline the first structure of the
work. The critical review shapes the work in terms of building up a frame including theories and processes by carefully collect different argumentations building them specifically on the research, being able to justify any usage of specific literature regarding the conducted analysis (Saunders, 2012, p. 72). The last review is seen to highlight the concrete chapters and discussions (Saunders, 2012, pp. 70-82).
Referring to Bryman & Bell’s (2011, p. 94) classification, we scrutinised literature using the systematic review in order to gather different studies taking them as a guide for our own research. Using the data base of i.e. Umea University library, google scholar, google books, and articles databases as EBSCOhost Business Source Premier, Jstors, Wiley Online Library, Elsevier ScienceDirect Journals by implementing broad keywords as i.e. investment strategies, profitability ratios, Novy-Marx + profitability, gross profitability, value investing, price multiples, portfolio management.
However, regarding the literature process explained by Saunders it can be argued, that our first review led us to different ideas for a relevant research topic. Nonetheless, the directive support of our supervisor and the critical review helped us to define the research. The critical literature review was also examined by constructing the research with theories using more specific keywords referring to Novy-Marx gross profit-to- asset, profitability ratio, gross profit, quality investing, value investing, time series analysis, gross profitability premium. To stress different argumentation points of the discussion more literature was consulted using keywords as i.e. financial statistics analysis, book-to-market interpretation, price multiples downturns, gross profitability downturns, cross-sectional returns behaviour.
The data itself can be derived either external or internal, whereas the latter can be defined as the access is rejected for other companies’ researchers. Internal data can often be described as highly accurate with beneficial information (Zikmund et al., 2012, p. 70). External data is the data, which can be perceived outside of the company of the concerned researcher (Bryman & Bell, 2011, p. 172). The data collection can be broadly differentiated into three classifications. Collecting new data can be difficult in terms of time and financial support. The familiarity with the data is high. Secondary analysis is based on already collected data. This analysis has benefits regarding the time consuming and financial terms. On the other hand, the researcher does not necessarily know the data, which can lead to mistakes regarding interpretation (Bryman & Bell, 2011, p. 313). Official statistics refers to data which is as the secondary data already collected but by an official institution in order to provide support for research. However, the data quality is also dependent on the collector. Thus, the researcher has no control about its quality. As it is official it often has a broad usage, is reliable and is sufficient for both, cross sectional and longitudinal studies (Bryman & Bell, 2011, p. 327).
We are using the data collected by Thomson and Reuters. We chose this data base as it is often referred as providing high quality data and it is available to us free of charge through our university. Furthermore similar studies, as e.g. “Rinne & Vähämma, 2011;
Elze, 2010; Chi et al., 2014” are also based on Thomson Reuters DataStream which
supports our decision to employ it as well.
2.8. Reliability, Replication and Validity
Reliability, Replication and validity are seen to assess the research’s goodness (Bryman
& Bell, 2011, p. 40). Reliability expresses the consistency of the findings. It refers to the correct choice of methods and measurements to produce a reproducible result. It is closely linked to the idea of replication, the accurateness of notes reflecting every step of the research processed. It aims at giving a third person the possibility to replicate the conducted research process and methods. Replication expresses the stableness of a research measured if it is repeatable and is therefore connected with the chosen approach, especially with the quantitative approach (Bryman & Bell, 2011, p. 41).
Validity is seen as the most important attribute as it inherits the evidence characteristic concerning integrity (Bryman & Bell, 2011, p. 41). Validity can be sub-divided into measurement validity, internal validity, external validity and ecological validity.
Measurement validity deals with the link between the concept and the measurement which has to precisely reflect the concept. Internal validity refers to the significant causality between the evidence and the measured. External validity is concerned with the results having a generalising effect on the examined area. Last but not least ecological validity expresses the usability of the findings on the common life of social actors (Bryman & Bell, 2011, pp. 41-43). In addition, Zikmund et al. (2012) require also the timeliness of a study, reflecting the relevance for a contemporary study (Zikmund et al., 2012, p. 1).
The reliability of our study is covered following the method choice and position of the previous chapters as well as the conduction in chapter four. We underpin our choice of methods with previous studies which have examined similar researches. Moreover, the data is derived from an official source, which indicates high quality data. The research is based on historical data implying that the data does not get adjusted over time leading to a high stability of findings. The measurements to calculate the results are generally accepted and applied also in other studies.
Nevertheless, we ensure the replicability by precisely note and present each step directed concerning this study. Furthermore, the data that we use is company data that are legally and officially available, which underlines replicability of the study. We strive for arguing regarding decision and thoughts, which come up during the process in order to give the reader the opportunity to follow and understand our implications.
Validity can be examined by contemplate chapter four and five of this paper. We aim
for validity of measurement by consulting similar studies evaluating a similar purpose
and derive applicable measurements. Further we review different statistical literature in
order to gather specific knowledge of possible measurements. The internal validity is
given as the results are directly derived from the given data. The External validity can
be viewed due to a generalised statement about similar markets as the examined
NASDAQ OMX Stockholm exchange market. It can be argued that external validity is
only given to a certain extent, as the results are only generalizable under the taken
assumptions. The ecological validity however, is difficult to analyse as not every social
actor might have an interest in the usage of the findings. Nevertheless, the research
contributes to all social actors concerned with e.g. investment strategies in small
exchange markets in the broad sense.
The timeliness stressed by Zikmund et al. (2012, p.21) is achieved by this paper referring to the recent development of the gross profitability ratio. The main paper of Novy-Marx introducing this specific ratio was published in the 2010 and his subsequent paper “Quality investing” was published in 2013. From then on it was subjected in some different studies, but as far as we know this paper is the first to examine the ratio referred to value strategies on the NASDAQ OMX Stockholm exchange market in the way we present it in this thesis. However, it has to be stressed, that the study is not only of current value but also consultable in later years, as this study generates results of a significant time period.
2.9. Ethical and Societal Issues
Ethical issues can evolve at different stages during the research process. They should not be involved due to their close linkage to integrity. It is referred to value and the treatment of other people and relation towards concerned actors. One can adhere to ethical codes of different associations (Bryman & Bell, 2011, pp. 120-121).
Due to different ethical perceptions of researchers there exists a huge difference in ethical considerations. In terms of a research this can be the usage of unofficial data, citation failures, or harm on the research object (Bryman & Bell, 2011, pp. 120-121).
Bryman & Bell (2011, p. 124) distinguish between several different ethic-based attitudes, which can lead to deferred outcomes. This is not further explained in order to not extent the scope of this thesis.
However, Bryman and Bell (2011, pp. 139) name i.e. deception and “data management”
to be considered ethically, which is crucial for our study. It is concerned with the different steps where data is involved to the concern of legitimation of usage and needs clarification of the legal ownership and rights of use. But also the presentation of the data is important to consider. On the other hand, deception deals with the incorrectness in reflecting e.g. the result or process due to i.e. self-interest.
Hence, deceptions as well as data management are crucial to consider making decisions
i.e. during the process and interpretation of our study. We pay attention to every step
along the process and statement of the results to not make mistakes by accident. Further
we conceive deception as ethically questionable and thus, we are not willingly
adulterate results or statements. We aim at accurately illustrate the steps along our
study, i.e. finding out if the gross profitability ratio of Novy-Marx can be applied on the
Swedish market. The application of models, analyses and interpretation is conduct to
our best knowledge and conviction. The legal possession and usage of the data is
supported through Umeå University having a license to draw data from Thomson and
3. Theoretical Framework and Literature Review
This chapter provides the theoretical background for the study. It explains the theories that are important to understand and to follow the conducted research. The investment strategies, that are used for the portfolio construction (see chapter 4.2.2.), are clarified, as well as the corresponding ratios. The chapter closes presenting previous research.
3.1. Efficient Markets Hypothesis and the Random Walk
Fama (1991, p. 1579) claims that the Efficient market hypothesis (EMH) is a very popular hypothesis among academics. A market is called efficient if a price of a security fully reflects available information (Fama, 1970, p. 383). Anything that might affect the market price, which is not known in the present or is a random change in the future, is defined as information or news (Clarke et al., 2001, p. 1). The market hypothesis assumes that it is not possible to predictably outperform the market in the long-term, as all relevant information is costless and available for all market participants. However, following the idea of the random walk it is possible to outperform the market by chance (Clarke et al., 2001, pp. 10-12). This is reflected by the implicated agreement on the already priced information regarding the today’s selling price and future’s selling price (Fama, 1970, p. 387). However, it further assumes that no transaction costs exist. In efficient markets stock prices follow a random walk, meaning that stock movements are random, thus, the price changes independently from the previous price movement.
Consequently, a price or its movement is not possible to predict.
Market efficiency is divided into three different forms. In literature a distinction is drawn between weak, semi-strong and strong market efficiency. Broadly, it inherits the idea that the degree of market efficiency increases by the level of information, which is included in a security’s price (Fama, 1970, p. 383).
A weak form of market efficiency exists when the current price for a security contains all stock market information e.g. as trading volume, historical prices, or short-term interest. Since the stock price at time t reflects all past information, the idea is rejected that the future price change is predictable by historical information. This is also in line with the assumption of the random walk theory, explained further down in this chapter (Bodie et al., 2011; Fama, 1970, p.383).
As an extension of the former the semi-strong form of market efficiency states, that prices instantly reflect all published information, such as announcements, annual reports or fundamental analyses, as well as all public market information. Hence, in a semi- strong efficient market it is not possible for an investor to beat the market by just having public information, because these are already fully reflected in the price (Bodie et al., 2011, p. 376; Fama, 1970, p.383).
The strong form on market efficiency states that the prices do not only reflect historical
information and publicly announced information, but also internal information of the
concerned company (Fama, 1970, p.383). Bodie et al. (2011, p. 376) argues that this
monopolistic information is just available to some investors or groups. To give an
example, one can describe that employees may have access to internal information of a
firm before they get published (Fama, 1970, p. 383, 415; Bodie et al., 2011, p. 376). It
can be argued, that insider traders use this information to adjust market prices.
However, insider trading is forbidden in most countries of the world (Arshadi, 1998, p.
70). Furthermore, it can be stated, that it is impossible that stock prices reflect the value of insider information (Fama, 1991, p. 1577). It is empirically proved, that insider possesses the ability to gain abnormal returns conducting insider trading (Seyhun, 1986, pp. 210-211). Hence, this fact partly rejects the strong form of the EMH. If the strong efficient market theory holds it does not matter which investment strategy is followed, the return would be equal and thus, the only way to beat the market is through experiencing luck.
Nevertheless, the price changes in efficient market follow a random walk as prices already contain all information, leaving no room for exploitation. The random walk theory assumes price changes to be independent of each other and follow the same distribution. In 1900 the French mathematician Bachelier first mentioned a phenomenon of random processes in his dissertation (Fama, 1970, p. 389). Kendall (1953, p. 13) found that stock prices at time t are independent from historical prices and follow a random walk (Kendall, 1953, p. 13). In other words, the prediction of future prices is not possible by looking at previous stock price movements (Fama, 1970, p. 388). That indicates that beating the market is not possible (Fama, 1970, p. 412).
Figure 2 - Upwards Sloping Random Walk.
Source: Statistics and Finance: An Introduction. Ruppert, 2004, p. 83