Value Investing: Predicting Long-Term Pro?tability Based on Fundamental Data

Value Investing: Predicting Long-term Pro? tability Based on Fundamental Data An Empirical Study in the Manufacturing Industry by Vital Schwander (05-609-136) Master’s Thesis supervised by Prof. Dr. Andreas Gruner University of St. Gallen May 23, 2011 Master in Law & Economics Abstract Warren Bu? ett (1992) classi? es the discussion about value and growth stocks as fuzzy thinking. With that statement, he argues that value investors must consider growth in their value calculations. This thesis shows in a ? rst step that growth is only valuable if the company enjoys a durable competitive advantage.By examining the fundamental characteristics of companies with a durable competitive advantage, this thesis intends in a second step to assess the predictability of long-term pro? tability. The DuPont Identity serves as framework for that purpose. The objects of this investigation are companies within the manufacturing industry (Primary SIC Code between 2000-3999) that were listed in t he United States between 1979 and 2009. The results show that companies with a durable competitive advantage exhibit speci? c characteristics in operating e? ciency, asset use e? ciency, and in the ability to meet short-term obligations.Furthermore, the thesis shows that long-term pro? tability, based on the investigated characteristics, is predictable to some extent. This thesis concludes by assembling the insights to a value strategy that is applied to manufacturing companies listed in Switzerland. The strategy exhibits an outstanding SMI-adj. compound annual growth rate of 13. 19% over a period of 17. 5 years. ii Acknowledgement I would like to express my gratitude to Prof. Dr. Andreas Gruner for supervising this thesis and his assistant Lucia Ehn for her conceptual advices. I have furthermore to thank Mr.Hans Ulrich Jost for giving me insight into the daily business of a value fund at UBS AG. My sister Daria introduced me to R and Latex. I want to thank her for her help and supp ort. I want to thank my great family who has been always supportive and motivating. Finally, I also would like to thank friends and colleagues for making life such an enjoyable experience. iii Contents 1 Introduction 1. 1 1. 2 Issues, Goals and Limitations . . . . . . . . . . . . . . . . . . . . . . . . . Structure and Empirical Approach . . . . . . . . . . . . . . . . . . . . . . 1 1 2 4 4 5 7 7 8 2 Value Investing—An Investment Paradigm 2. 2. 2 2. 3 The Origin of Value Investing . . . . . . . . . . . . . . . . . . . . . . . . . Value and Other Investors . . . . . . . . . . . . . . . . . . . . . . . . . . . Four Value Strategies by Illustration . . . . . . . . . . . . . . . . . . . . . 2. 3. 1 2. 3. 2 2. 3. 3 2. 3. 4 2. 4 2. 5 Piotroski’s F_Score . . . . . . . . . . . . . . . . . . . . . . . . . . Walter and Edwin Schloss . . . . . . . . . . . . . . . . . . . . . . . Warren Bu? ett . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 UBS EMU Value Focus Fund . . . . . . . . . . . . . . . . . . . . . 12 Value vs Growth—Fuzzy Thinking! . . . . . . . . . . . . . . . . . . . . . 13 Value Anomaly . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 2. 5. 1 2. 5. 2 2. 5. 3 Behavioral Approach . . . . . . . . . . . . . . . . . . . . . . . . . . 14 Risk-based Approach . . . . . . . . . . . . . . . . . . . . . . . . . . 14 Competitive Advantage Based Approach . . . . . . . . . . . . . . . 16 17 3 Literature Review 3. 1 3. 2 3. 3 Competitive Advantage . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 Pro? tability Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . 20 Research Gap and General Approach . . . . . . . . . . . . . . . . . . . . 21 22 4 Analysis of Long-term Pro? tability 4. 1 4. 2 Data Sample . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 Analysis of Return on Equity Measure . . . . . . . . . . . . . . . . . . . . 26 4. 2. 1 Superior Performers . . . . . . . . . . . . . . . . . . . . . . . . . . 26 iv 4. 2. 2 4. 2. 3 4. 3 4. 4 4. 5 4. 6 Analysis of Performance Persistence . . . . . . . . . . . . . . . . . 28 Analysis of SPP Deciles in respect of ROE . . . . . . . . . . . . . . 30 Analysis of SPP Deciles in respect of other Financial Measures . . . . . 33 Predictability of Long-term Pro? tability . . . . . . . . . . . . . . . . . . . 41 Discussion of the Interim Results . . . . . . . . . . . . . . . . . . . . . . . 43 Market Analysis 4. 6. 1 4. 6. 2 4. 6. 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 Subdivision-speci? c Market Analysis . . . . . . . . . . . . . . . . . 45 Analysis of SPP Deciles in respect of Market Multiples . . . . . . . 45 Market Performance Analysis . . . . . . . . . . . . . . . . . . . . . 46 48 5 Value Strategy 5. 1 5. 1. 1 5. 1. 2 5. 1. 3 5. 2 Strategy Composition . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48 Sample Descriptives and Strategy Composition . . . . . . . . . . . 48 Portfol io Formation . . . . . . . . . . . . . . . . . . . . . . . . . . 49 Performance Measurement . . . . . . . . . . . . . . . . . . . . . . . 49 Portfolio Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 53 6 Conclusion and Further Research 6. 1 6. 2 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 Further Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 56 i v x xv Bibliography A Data Input B Financial Measures C Subdivisions D Market Analysis List of Tables 4. 1 4. 2 4. 3 4. 4 4. 5 4. 6 4. 7 4. 8 4. 9 5. 1 COMPUSTAT Items . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 . . . . . . . . . . . . . . . . . . . . . . . . . . 25 Distribution of Firm Years Distribution of Superior Performance Years . . . . . . . . . . . . . . . . . 27 Probability Distribution of Superior Performance Persistence . . . . . . . 29 ROE Distribution for each SPP Decile . . . . . . . . . . . . . . . . . . . . 31 RO E Distribution for each SPP Decile (Subdivision-adjusted) . . . . . . . 32 Financial Measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 . . . . . . . . . . . . . . . . . . 47 Predictability of Future Pro? tability . . . . . . . . . . . . . . . . . . . . . 42 Market Performance for each SPP Decile Portfolio Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 . . . . . . . . . . . . . . . . . . . . . . . . i ii v A. 1 Data Input for US Companies A. 2 Data Input for Swiss Companies . . . . . . . . . . . . . . . . . . . . . . . B. 1 Calculation of the Financial Measures . . . . . . . . . . . . . . . . . . . . B. 2 SPP Deciles (Subdivision-adjusted) regarding Financial Measures . . . . vii x xi C. Overview of Subdivision . . . . . . . . . . . . . . . . . . . . . . . . . . . . C. 2 Subdivision Comparison regarding ROE . . . . . . . . . . . . . . . . . . . C. 4 Composition of SPP Deciles regarding Subdivisions C. 3 Subdivision Distribution in respec t of SPP Deciles . . . . . . . . . . . . . xii . . . . . . . . . . . . xiii D. 1 Average Price-Earnings Ratio per Subdivision . . . . . . . . . . . . . . . . xvi D. 2 Average Book-to-Market Ratio per Subdivision . . . . . . . . . . . . . . . xvii D. 3 Average Price-Earning Ratio per SPP Decile D. 4 Average Book-to-Market Ratio per SPP Decile . . . . . . . . . . . . . . . xviii . . . . . . . . . . . . . . . xix vi List of Figures 3. 1 4. 1 4. 2 4. 3 5. 1 Three Slices of Value . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 . . . . . . . . . . . . . . . . . . . . . . . 33 . . . . . . . . . . . . . . . . . . 47 Mean ROE for each SPP Decile SPP Deciles in terms of Financial Measures . . . . . . . . . . . . . . . . . 40 Market Performance for each SPP Decile Performance of the Value Strategy . . . . . . . . . . . . . . . . . . . . . . 51 . . . . . . . . . . . . . . . . . . . . . . . . . . . xiv C. 1 Subdivision Distribution vii List of Abbreviations vg. B/M CAP CAGR CAPM COG S DA EBITDA etc. e. g. EV FCF IE i. e. IPO LT p. a. P/E ROA ROE SGA SMI ST US average Book-to-Market Competitive Advantage Period Compound Annual Growth Rate Capital Asset Pricing Model Cost of Goods Sold Depreciation and Amortization Earnings before Interest, Taxes, Depreciation and Amortization et cetera exempli gratio – for example Enterprise Value Free Cash Flow Interest Expense id est – that is Initial Public O? ering Long-term per annum Price-Earnings Return on Assets Return on Equity Selling, General, and Administration Swiss Market Index Short-term United States iii Chapter 1 Introduction 1. 1 Issues, Goals and Limitations Every investor is looking to buy low and sell high. This does not yet characterize a value investor. Although value investing has become a widely used term, it has been stamped in particular by a small group of academics. They associate stock-speci? c fundamentals such as a low P/E ratio, low cash-? ow-to-price ratio, and high B/M ratio to va lue stocks. These stock-speci? c fundamentals have become characterizing for value investing and embody the basis for many research studies about value investing (see Damodaran, 2011).For example, Piotroski (2000) developed the F_Score to separate losers from winners among value stocks (i. e. high B/M-stocks). On the other hand, research has been conducted on growth stocks (i. e. high P/E ratio, high cash-? ow-to-price ratio, and low B/M ratio). Mohanram (2005) developed the GSCORE to separate losers from winners among growth stocks, for instance. As a consequence, many investors feel compelled to decide between value and growth stocks. However, in the heated discussion it is often ignored that growth has an impact on the value of a company.This impact of growth varies according to the particular company from negligible to very important, and its impact can be negative as well as positive. Growth is valuable in particular if a company enjoys a durable competitive advantage and remai ns very pro? table over a long period of time. There are many books about the competitive advantage (e. g. Porter, 1998; Shapiro, 1999). However, it has never been discussed related to value investing. Only Mauboussin and Johnson (1997) have raised a discussion about the competitive advantage period within the valuation process of stocks.They point out in their paper „Competitive Advantage Period: The Neglected Value Driver† that the persistence of competitive 1 advantage has a huge impact on the value of a ? rm. Yet there is little literature on this topic (see Fritz, 2008) and the bulk of academics as well as practitioners still rely mainly on the di? erentiation between value and growth stocks. This thesis gives priority to the competitive advantage, though, and intends to lay the groundwork for valuing competitive advantage. It is important to understand how a competitive advantage can be captured and if it is possible to predict long-term pro? ability, before starti ng to value the growth potential of a company. Hence, the aim of the thesis is con? ned to the predictability of long-term pro? tability and does not intend to value the competitive advantage as such. The ? rst question that arises in this context is whether it is possible that a company can exhibit long-term pro? tability. The answer to this question is of interest, as most economists maintain the contrary. According to economic theory, pro? tability is mean reverting in a competitive environment (Chan, Karceski and Lakonishiok, 2003). However, reality teaches us the contrary every day.Mircrosoft’s products, for instance, are everything else but innovative. Nevertheless, the company earns excessive returns for decades, and so do others like The Coca-Cola Company. Thus, this thesis investigates the possibility that a company is able to sustain its competitive advantage over several years. Thereupon, the second issue addresses whether companies with a durable competitive advan tage exhibit stock-speci? c fundamental characteristics. Therefore, the DuPont Identity serves as framework. The companies are classi? ed into deciles in terms of pro? tability (i. e.ROE) and persistence. Upon this, the companies are tested for the characteristics regarding various measures, which are derived mainly from the DuPont Identity. All companies that are objects of the investigation are listed in the United States and constrained to manufacturing companies only. The third question addresses whether it is possible to identify companies with a durable competitive advantage based on the observed characteristics. Finally, a simple strategy is composed that implements the investigated characteristics of companies with a durable competitive advantage.The strategy is conducted on manufacturing companies that are listed on a Swiss stock exchange. 1. 2 Structure and Empirical Approach The present thesis is structured in mainly four parts: Chapter 2 reviews literature on value inves ting and points out the broad range of value strategies by the mean of four examples. The reader shall gain an overview of value investing (i. e. the origin of value investing, dissociation from other investors, and current value discussion). Additionally, 2 this chapter shall point out the link between value investing and the competitive advantage period.Chapter 3 contains a literature review about competitive advantage, pro? tability measures and the persistence of pro? tability. Moreover, chapter 3 shows the research gap as well as the general approach to ? ll this gap. The empirical part in chapter 4 deals mainly with three issues: (1) persistence of superior performance, (2) characteristics of companies with a competitive advantage, and (3) predictability of future long-term pro? tability. Finally, in chapter 5 a value strategy will be composed that builds on the insights of chapter 4. 3 Chapter 2 Value Investing—An Investment Paradigm 2. 1 The Origin of Value InvestingV alue investing is an investment paradigm that derives its origin from the ideas on investment and speculation subsequently developed and re? ned by Benjamin Graham and David Dodd through various editions of their famous book Security Analysis. Starting in 1928, Graham began to teach a course on security analysis at Columbia University. The book emerged from that course, and appeared in 1934. Graham and Dodd mainly summed up lessons learned from the previous economic crisis in 1929 and provided readers with inevitable principles and techniques by focusing on the analysis of fundamental ? gures to estimate the value lying behind securities.By publishing the ? rst professional book about investing, they laid the foundation of value investing. In 1949, Graham published his second book, The Intelligent Investor, which was described by Warren Bu? ett (Graham, 2003) as „by far the best book on investing ever written. † It contains mainly the same ideas as in its predecessor Sec urity Analysis, but focuses more on the emotional aspects of stock markets, rather than on analyzing techniques. The techniques to determine investment opportunities that Graham and Dodd have developed are based on two fundamental assumptions about the market: 1.Market prices of securities are sometimes subject to signi? cant and unforeseeable movements. 2. As opposed to the e? cient market hypothesis, which assumes that all stocks are correctly priced by the market at any one point in time, market prices of some 4 securities deviate from their intrinsic values from time to time despite the fact that their underlying economic values do not justify such signi? cant deviation. Hence, an intelligent investment is characterized as paying less for a security then its intrinsic value. Paying more for a stock than its intrinsic value in the hope that it can be sold for a higher price is speculative.In other words, an intelligent investor should not attempt to forecast future stock market m ovements; instead, such movements provide opportunities to purchase undervalued stocks. Moreover, investors are encouraged to purchase securities only when the market price is su? ciently below its intrinsic value. Graham (2003) referred to this signi? cant gap between price and intrinsic value as the margin of safety, and quali? ed it as central concept of investment. In practice, investors lay down di? erent margins of safety that are appropriate to their fundamental analysis. A super? ial analysis requires a higher margin of safety than a deep and broad analysis. Additionally, market conditions as well as the sizes of funds gives reason for di? erent margins of safety. Bu? ett states in his letters to the shareholders of Berkshire Hathaway, Inc. in 1992: „We have seen cause to make only one change in this creed: Because of both market conditions and our size, we now substitute ,an attractive price’ for ,a very attractive price’ (p. 12). † Yee (2008) sugg ests a margin of safety between 10% and 25% of the share price. Larger margins are justi? ed for especially risky stocks.Accordingly, the margin of safety is not a rigid safety net but rather a ? exible net with meshes, which must be properly adjusted to the speci? c needs and conditions from time to time. 2. 2 Value and Other Investors Classic value investors—in the sense of Graham and Dodd—are rare. Every investor is looking to buy low and sell high, but what exactly di? erentiates a real value investor from all the other investors? According to Greenwald, Kahn, Sonkin, van Biema (2001), investors can be di? erentiated into two main categories. The ? rst category pays no attention to fundamental analysis.Instead, these investors analyze charts; in particular they construct charts to represent trading data (e. g. price movements and volume ? gures). In other words, they intend to predict future price movements referring to previous events regardless of its fundamental value (pp. 5-6). Graham and Dodd qualify these investments as highly speculative. 5 Although the second category focuses admittedly on fundamental analysis, Graham and Dood value investors are still a minuscule minority. Greenwald, Kahn, Sonkin, van Biema (2001) divide these fundamentalists into those who ocus on macroeconomics and those who deal with the microeconomics of securities. Macro-fundamentalists often pursue a top-down approach by considering ? rst broad economic factors such as interest rate, in? ation rate, exchange rate, unemployment rate, and the like. They forecast economic trends on a broad national or even worldwide basis. Upon this, they decide whether a group or even a speci? c security is a? ected by this trend. They do not calculate the value of individual securities, though. In particular, they monitor policy makers, such as the central bank, and try then to determine the impact on a speci? industry or group of securities. As any other investor, they attempt to forecast price movements before other investors recognize them and subsequently buy low and sell high, but they do not calculate the intrinsic value of an individual security directly (pp. 6-7). Graham and Dodd originally established value investing as a comprehensive analysis of securities in order to estimate the intrinsic value as accurately as possible, but in the group of micro-fundamentalists, traditional value investors are still a minority.According to Greenwald, Kahn, Sonkin, van Biema (2001), a more common approach takes the current price of a stock as the point of departure. These investors analyze the history of a security, considering how the stock price was in? uenced by changes in the underlying economic factors. In a second step they then attempt to predict the probability and impact of such changes in order to forecast future development of the speci? c security. These kind of investors often forecast future earnings or free cash ? ows. If they ? d that their pre dictions are more optimistic than the market’s expectation, they buy the security; if they ? nd that the market’s overall expectation is to high compared to their forecast they sell the security (p. 7). Indeed, most value investors—in the sense of Graham and Dodd—start their analysis from the bottom up by calculating ? rst the intrinsic value of a ? rm and subsequently they estimate the macroeconomic exposure of the ? rm—similar to the micro-fundamentalists. Although there are some similarities, Graham and Dodd value investors distinguish themselves from micro-fundamentalists in many ways.Greenwald, Kahn, Sonkin, van Biema (2001) mention two reasons why most micro-fundamentalists are not value investors: First, they focus on prior and anticipated changes in prices, and not on the level of prices relative to underlying values. The second and even more decisive di? erence is the absence of a margin of safety to safeguard investors from unpredictable market movements (pp. 7-8). Accordingly, a true value investor in the classical sense is one whose point of de6 parture is the fundamental data of a company. Although macro-economic factors play a signi? cant role in the analysis, they are of secondary importance.Furthermore, this investor does not predict future developments of key factors that cause price changes. Instead, a classic value investor values a company based on current fundamentals and buys a security at a bargain price. In the following section, four value strategies are outlined in order to give an idea by the way of illustration. 2. 3 Four Value Strategies by Illustration The range of value strategies is broad enough that it makes it impossible to sum up all of them. Thus, the following selection intends to show the large variety of aspects that these strategies characterize. These aspects range from fundamental analysis only (e. . Piotroski) to more sophisticated investigation of companies (e. g. Bu? ett), from con centrated portfolios (e. g. UBS EMU Value Focus Fund) to diversi? ed portfolios (e. g. Schloss). 2. 3. 1 Piotroski’s F_Score Piotroski started his career as a professor at the University of Chicago Graduate School, and since 2007 he has taught accounting at the Stanford University Graduate School of Business. In April 2000, Piotroski published a paper in the Journal of Accounting research titled „Value Investing: The Use of Historical Financial Statement Information to Separate Winners from Losers. In this paper, Piotroski classi? es distressed companies in winners and losers by means of nine fundamental criteria. Four criteria (ROA, ? ROA, CFO, and ACCRUAL) re? ect the pro? tability, three criteria (? LEVER, ? LIQUID, future debt obligations, and two criteria (? MARGIN, and ? TURN) measure changes F_Score is composed as follows: F _Score =F _ROA + F _ ? ROA + F _CF O + F _ACCRU AL + F _ ? LIQU ID + EQ_OF F ER and EQ_OFFER) measure changes in capital structure and the ? rm’s ability to meet in the e? ciency of the ? rm’s operations (Piotroski, 2000, pp. 10-14).The equation of the + F _ ? M ARGIN + F _ ? T U RN + F _ ? LEV ER (2. 1) where a low F_Score signals a ? rm with less recovering potential and a high score indicates the ? rm as having mostly good prospects to recover. If a company ful? lls a criteria, 7 the F_criteria equals 1, otherwise 0. With that, Piotroski translates the criteria into binary signals. The sum of all F_criteria subsequently leads to the F_Score, which can range from a low of 0 to a high of 9. Due to the fact that it is very di? cult to obtain the maximal score, companies with a minimum score of 8 will be classi? d as high F_Score whereas as companies with a score of 0 or 1 are classi? ed as low F_Score (Piotroski, 2000, pp. 14-18). Piotroski (2000) reevaluates the stocks every year and decides whether a stock belongs to the losers or to the winners. Finally, the investment strategy buys high F_Score and sel ls short the low F_Score. This simple strategy generates over two decades an astonishing 23% average annual return. It appears that the strategy is also robust in crisis. In 2008, the American Association of Individual Investors tested the strategy among 50 other investment strategies.With a performance through to the end of 2008 of 32. 6%, it was not only the only stock strategy that would have generated positive returns but has also outperformed the median performance (-41. 7%) of all tested strategies by far (Thorp, 2009). Due to the fact that the portfolio is construed each year on actual data, it is often the case that the portfolio is turning over correspondingly. Once a ? rm is recovering and the market has recognized the improvements the B/M ratio increases and the stock does not appear any more on the screen, although the company has even more growth potential. That is why many ? ms remain no longer than one or two years in the portfolio. Admittedly, buying winners and shor t-selling losers is one big advantage of the strategy. Companies that are classi? ed as losers may transform in a subsequent period from a low F_Score to a high F_Score ? rm. Therefore, the strategy makes double use of a company’s development or business cycle. But the strategy also implies a disadvantage; why should an investor sell an excellent business that bought at a bargain price? Based on a competitive advantage, the business could thrive to a superstar and yield high returns on the initial investment.A top manager also keeps the good business also when others o? er more than its current value because the manager knows that the business will contribute also in the future to the ? rm and its shareholders. 2. 3. 2 Walter and Edwin Schloss Walter Schloss and his son Edwin are very conservative value investors whose motto is to keep things simple and cheap. Walter Schloss attended a course of Graham’s and worked for the Graham-Newman Partnership until 1955. Afterwar d, he ran his own investment ? rm and in 1973 his son Edwin joined the partnership. From the formation of the limited 8 artnership until 2000, the Schloss have provided their investors an annual compound return of 15. 3%. They outperformed the S&P Industrial Index by 4. 2% annually. In other words, they have created a return of 66,200% while the S&P Industrial Index performed 11,800% (Greenwald, Kahn, Sonkin, van Biema, 2001, p. 263). Walter Schloss has been titled by Warren Bu? ett as „superinvestor† (Forbes, 2008). What distinguishes the Schlosses from other value investors is their simple, and almost rudimentary method choosing stocks. They are among the few investors that stick to the principles of the father of value investing.Like Graham, they seek for stocks that are priced lower than their working capital (net assets minus current liabilities). They start their investigation by putting their feelers out to stocks that are unloved, distressed, and unheeded from ot her investors. Most of these stocks are in a downward trend either by a rapid plunge or a continually decreasing price. The longer the company has gone through such hard times, the more they call the Schloss’s attention. Once they have invested in such a unloved stock they hold it on average for four to ? ve years until the stock has recovered. Sometimes they also sell a stock earlier when they ? d a better opportunity (Greenwald, Kahn, Sonkin, van Biema, 2001, pp. 266-269). Edwin Schloss focuses on asset values, but is also willing to buy a company that has a strong earnings power. Greenwald, Kahn, Sonkin, van Biema (2001) describe the investment philosophy of Edwin Schloss as follows: „Edwin Schloss pays attention to asset values, but he is more willing to look at a company’s earnings power. He does want some asset protection. If he ? nds a cheap stock based on normalized earnings power, he generally will not consider it if he has to pay more than three times b ook value. [†¦ Depending on his estimate of what the companies can earn, Edwin may still ? nd the stock cheap enough to buy (p. 268). † Although Edwin pursues a more liberal value approach by taking the earnings power value into account, he is still very conservative. Both father and son do not include in their valuation process other than fundamental data. In their analysis, they rely entirely on annual and quarterly reports—they keep things simple but with a relatively high margin of safety. The diversi? cation of their portfolio also varies. They do not determine a threshold in advance to which they stick.Similar to Warren Bu? ett, their approach leads them to industries, which are not exposed too much to rapid changes that can undermine the value of these stocks (Greenwald, Kahn, Sonkin, van Biema, 2001, p. 269). 9 2. 3. 3 Warren Bu? ett Warren Bu? ett, who is doubtless the most famous student of Graham and one of the most successful investors, too, pursues a s imple strategy, which is complex and di? cult in its execution. Bu? ett started his career in Graham’s investment ? rm. In 1964, he then bought shares of Berkshire, when its book value per share was $19. 46 and its intrinsic value even lower (Bu? tt and Cunningham, 1997, p. 6). In the period from 1964 to 2009, book value per share increased at an annual compound rate of 20. 3% that is an overall gain of 434,057 %. Adjusted by the S&P with dividends included, Berkshire has a compound annual growth rate of 11%. During the period, Berkshire reported only twice a negative change in book value—in 2001 and 2008—compared to the S&P that incurred during the same period eight negative results (Bu? ett, 2009, p. 2). Unlike other investors, Bu? ett feels obliged to share his knowledge that he gained mainly from Graham.Moreover, and opposed to the bulk of successful investors, he teaches his wisdom to the world of investors—and those who are interested in his activit y— by an annual letter to the shareholders of Berkshire Hathaway, Inc. To attain this knowledge it is not necessary to buy a share of Berkshire Hathaway, Inc. —which costs currently over $125,000, nor is it necessary to pay any money for it. Bu? ett gives access to his letter on the Berkshire’s website for free. Additionally, in the book called The Essays Of Warren Bu? ett—Lessons For Corporate America, Cunningham organizes the information in Bu? tt’s letters in a thematic way. This book is also accessible online and can be downloaded for free. Bu? ett is aware that he creates potential investment competitors by passing his wisdom to everyone but imitating Bu? ett’s strategy is everything but simple. His explanations are logical and easy to understand, but the execution requires much experience and a distinctive comprehension of the industry and costumer behavior. In contrast to what Piotroski and other academics and money managers postulate , Bu? ett buys not only high B/M stocks. This amazes readers in many ways. In particular, because Bu? tt refers in several passages of his letters to Graham’s conception. It also contradicts the conceptions of most academics, which assign a high B/M ratio to value stocks. Nonetheless, Bu? ett puts emphasis not only on the book value of a company but more on the competitive advantage that a company enjoys. Like Graham, he is looking primarily for very cheap businesses, which are traded far under their intrinsic values. As opposed to Graham, Bu? ett buys not every stock that Mr. Market o? ers him for a bargain price. Additionally, he seeks for businesses with a high competitive advantage.While most ? rms in Graham’s portfolio are distressed, Graham diversi? es the risk. Bu? ett, on the other hand, holds that an investor should not buy second-class stocks 10 in the hope that they will recover. The awareness of less investment opportunities does not bother Bu? ett; au cont raire, he avoids purchases that he will regret later. According to him, every transaction that is based on a wrong decision is unnecessary, and thus, to be avoided. One could say that transaction costs (e. g. trading costs) are tiny, that they carry no weight. But what most people disregard are taxes.With every transaction, book value is going to be reevaluated and governments levy taxes on the new value. Holding a share does not cause any taxes, as long as the investment will be sold. Therewith, Bu? ett did not pay taxes as much as his colleagues that trade frequently. Either way, Bu? ett’s preferred holding period is forever. This strategy particularly bene? ts private investors that have bought stocks of Berkshire Hathaway. At least in Switzerland, the government does not impose taxes on capital gains. In the shareholder letter from 1992, Bu? tt breaks his strategy down to a few cornerstones of the valuation process: „We select our marketable equity securities in muc h the way we would evaluate a business for acquisition in its entirety. We want the business to be one (a) that we can understand; (b) with favorable long-term prospects; (c) operated by honest and competent people; and (d) available at a very attractive price (p. 12). † First, Bu? ett never buys a business that he does not understand entirely. This requires a full comprehension about the industry such as competitors, value chain, costumers, and so on. For this reason, Bu? tt avoids industries with a high rate of change (e. g. technology industry). The second criterion that a business must live up to is a competitive advantage. Preferably, he is looking for businesses that have potential to improve their competitive positions within the industry. Third, but less important, Buffett is looking for competent management. It is less important, because according to him a company with a durable competitive advantage can even operate with ordinary managers and generate extraordinary r eturns (Bu? ett and Cunningham, 1997, p. 21). Finally, a margin of safety prevents Warren Bu? tt from mistakes or unforeseeable developments. It seems that soft factors play an important role for him in the valuation process. Correspondingly, fundamental analysis is only half the battle. The following quote from Warren Bu? ett in the context of the hostile takeover of RJR Nabisco outlines the kind of business Bu? ett likes: „I’ll tell you why I like the cigarette business. It costs a penny to make. Sell it for a dollar. It’s addictive. And there’s fantastic brand loyalty (Burrough and Helyar, 1991, p. 218). † 11 For this reason, Bu? ett also accepts businesses that do not always have a high B/M ratio.Moreover, he seeks for businesses that have potential for improvements and buys them at a relative bargain price in the hope the business remains its advantage and yields high returns in the future. 2. 3. 4 UBS EMU Value Focus Fund The UBS EMU Value Focu s Fund is a highly concentrated and actively managed European equity fund, which holds maximally ten stocks, where each has an initial weight of 10%. The investment process is divided into seven steps (Screening process; Short list; Pre due diligence; Full due diligence; Watch list; Entry, increase/reduce position; and Exit).First, the stock universe is screened by a quantitative approach (EV/EBITDA, P/E, B/M, FCF yield) and by a qualitative approach. Second, in the due diligence process the team meets the management of the target company, they compare the company within the peer group, and determine the fair value and entry level. The team gives particular importance to the within-industry comparisons and a margin of safety of 30%. After the stock is over the due diligence, the stock is deposited on the watch list until the entry level is reached. The stock remains in the portfolio until the stock has recovered and the calculated air value is reached and the weight of the stock is less than 15% of the portfolio. If there is a more promising investment opportunity, a position will be changed. Based on the high portfolio concentration, a sector limitation makes sure that stocks which are stemming from the same sector do not surpass the threshold of 33%. If a stock’s price plunges after its purchase more than 15%, the management also pulls the trigger for safety reasons and sells the stock (UBS, 2010). The strategy of the UBS EMU Value Focus Fund equals in some aspects Warren Bu? ett’s strategy.Both distinguish themselves from Piotroski’s and Schlosser’s strategy insofar as they include a due diligence process that goes beyond a fundamental analysis (e. g. valuation of the management). Furthermore, both strategies do not strive for diversi? cation, although the UBS EMU Value Focus Fund includes some risk management factors that compel the management to exit in certain circumstances. Warren Bu? ett, on the other hand, restricts himself by avoiding complex businesses. The two strategies also di? er insofar as the UBS EMU Value Focus Fund has a relatively short investment horizon of 18 months, whereas Bu? tt holds a stock over decades. 12 2. 4 Value vs Growth—Fuzzy Thinking! Although there is a broad variety of value strategies, it seems that the discussion about value investing leaves little room for interpretation. Nowadays, the bulk of academics di? erentiate between value and growth (glamour) stocks. They ? nd that stock-speci? c fundamental attributes such as a low P/E ratio (Basu, 1977; Ja? e, Keim, and Wester? eld, 1989), low cash-? ow-to-price ratio (Chan, Hamao, and Lakonishok, 1991), and high B/M ratio (Rosenberg, Reid, and Lanstein, 1985; Fama and French, 1992) earn substantially higher returns than glamour stocks.Hence, often one feels compelled to decide between value investing and growth investing. In particular, academic work has upheld the distinction, and thus, has had a strong impact on inv estment professionals. Furthermore, academic research developed style-speci? c benchmarks (Chan and Lakonishok, 2004, p. 71). In that sense, value stocks are referred to a high B/M ratio, a low P/E ratio and a high dividend yield, whereas opposite characteristics—a low B/M ratio, a high P/E ratio and a low dividend yield—are assigned to growth stocks. Some professional investment managers even see a mix of the two approaches as a smart cross-dressing.Among others, Warren Bu? ett labels this classi? cation as fuzzy thinking. Bu? ett argues that growth is always a component in the calculation of value. Nonetheless, he does not neglect that the importance of the growth component varies from negligible to very important and its impact can be positive as well as negative. Thus, a low B/M ratio, a high P/E ratio, and a low dividend yield is not per se inconsistent with „value† purchases. Business growth has often a positive impact on value but tells us little abo ut the intrinsic value of growth (Bu? ett, 1992, p. 12). All growth is not created equal, and thus must be di? erentiated.There is also value-destroying growth, which is not worth a penny. Bu? ett goes even further and scrutinizes the term value investing as such. According to him, the term is redundant because investing implies to pay less then the value of something (Bu? ett and Cunningham, 1998, p. 85). The origin of this fuzzy thinking constitutes the value anomaly that will be discussed in the following section. 2. 5 Value Anomaly Already Graham and Dodd (2008) hint at the discrepancy between market price and intrinsic value and the fact that the market often underestimates value stocks. This mispricing is called in the literature Value Anomaly.In the following section three explanations are outlined: i) a behavioral approach, ii) a risk-based approach, and iii) a competitive advantage based approach. 13 2. 5. 1 Behavioral Approach According to Graham and Dodd (2008), the irrat ional behavior of market participants can drive the price of a security in the wrong direction. As Graham outlined in his book the Intelligent Investor, emotions take part in the participant’s decisions, thus he rejects the E? cient-Market Hypothesis as well as the assumption of Homo Oeconomicus. Market participants are swayed either by positive emotions pushing up prices, or uncertainty and ? rce emotions cause a decline in prices. In general, both results in ine? cient and undesirable market upshots. De Bondt and Thaler (1985) already ? nd evidence that markets overact to unexpected and dramatic news events. Moreover, contagion ampli? es this process of counter-productive behavior, taking a central part of the game, especially in crisis when panics gain the upper hand and investors disinvest despite of existing reasons to act to the contrary. 1 Not only irrational behavior induces a discrepancy between market prices and intrinsic value. Discrepancies can also result from ? ms of little interest, and thus, small liquidity. In particular, small companies fall through the screening raster of professional investors. Once a professional investor manages a fund of a certain size, small investments are out of range. First, small companies are like gold dust, as a consequence thereof di? cult to ? nd, and second, the monitoring costs come along with the number of investments, which makes such companies unappealing. 2. 5. 2 Risk-based Approach Whereas Graham showed that behavioral aspects distort markets and cause a gap between intrinsic value and market value, many academics hold that the di? rence does not necessarily contradict the e? cient-market hypothesis. Some argue that higher returns simply compensate higher risk (Fama and French, 1994). As basis of this argumentation line served the Capital Asset Pricing Model (CAPM), which was developed independently by Sharpe (1964) and Linter (1965) in the 60’s based on Markovitz’s portfolio theory. The model shows the coherence between the expected return of individual securities and systematic risk (market risk). Whereby ? of a security is a parameter describing the relation of its return with that of the overall market.The equation of the CAPM can be summarized as follows: 1 Cella, Ellul, and Giannetti (2010) write in their paper about „Investors’ Horizon and the Ampli? cation of Market Shocks† that stocks which are held in a large part by short-term investors are more likely to plunge under their intrinsic value. They also instance that fund managers often follow restrictions, which do not lead to optimal purchases or sales. 14 E(Ri ) = Rf + ? i (E(Rm ) ? Rf ) (2. 2) where E(Ri ) is the expected return of a speci? c asset, Rf is the risk-free return rate, and E(Rm ) is the expected return of the market.Already Rosenberg, Reid, and Lanstein (1985) give rise to the assumption that the CAPM can not fully explain the correlation between expected returns and t he risk of an individual security. As a one factor model implies, the CAPM oversimpli? es the complex market. Therefore, Fama and French (1992) introduced a three-factor model that is an extension of the CAPM. Basically, they improved the CAPM by adding two more factors: (i) they distinguished between high and low B/M ratio, and (ii) classi? ed stocks according to market capitalization (price per stock times number of shares outstanding).The equation of the extended CAPM can be summarized as following: r = Rf + (Km ? Rf ) + bs ? SM B + bv ? HM L + ? (2. 3) where Rf is the risk-free return rate, Km is the return of the entire stock market, SM B (small minus big) is the di? erence between small and big ? rms according to their market capitalization, HM L (high minus low) is the di? erence between high and low B/M ? rms, bs is the corresponding coe? cient to SM B, and bv is the corresponding coe? cient to HM L. Based on this, Fama and French (1992) argue that high B/M ? rms’ pr ospects are judged relative poorly to ? ms with low B/M ratios. As already postulated by Chan and Chen (1991), Fama and French also interpret high B/M ? rms as ? nancially distressed (see also Piotroski, 2000). They adduce the explanation that a high B/M ratio inheres in a relatively high ? rm’s market leverage compared to its book leverage. Furthermore, they ? nd that during some periods (at least ? ve years) low B/M ? rms remain more pro? table than high B/M ? rms. Fama and French (1992) argue that more risk is inherent with a higher B/M ratio. In other words, value stocks are riskier than „glamour† stocks. Opposed to this, Gri? and Lemmon (2002) show that large returns of high B/M ? rms are inconsistent with a risk-based explanation. Arshanapalli et al. (1998) show 15 that value stocks generally have a risk-adjusted performance superior to that of growth stocks (p. 23). Thus, the value anomaly can be traced back to a mispricing of stocks due to overly optimisti c valuations of „glamour† ? rms. Once this mispricing is revealed, these ? rms earn negative excess returns. According to Chan and Lakonishok (2004), investors, in particular professional investment managers, focus their attention on apparent „glamour† stocks while stock prices of high B/M ? ms plunge under their fundamental value. Hence, investing in high B/M ? rms is likely to be a rewarded long-term investment strategy (p. 85). Moreover, Anderson and Smith (2006) ? nd that a portfolio of the most admirable companies substantially outperforms the market, and thus contradicts the e? cient market hypothesis. As a consequence, the risk-based explanation has lost many of its supporters over the last years and the value anomaly remained unexplained. 2. 5. 3 Competitive Advantage Based ApproachAlthough it is probably the closest explanation, academics rarely make the competitive advantage of a company accountable for the superior performance and excess returns of a company. According to them, competitive advantages must theoretically fade away. But in reality this is not always the case. New academic research indicates that the risk driver refers more to the riskiness of losing the competitive advantage (Mauboussin and Johnson, 1997; Greenwald, Kahn, Sonkin, and van Biema, 2001). This could be the case if new competitors enter the market and/or in industries where the rate of technology changes is high.On the one hand, new technologies open up new opportunities for existing players, but on the other hand, they also carry the risk that entrants come up with new products and technologies that force existing players to keep up with the changes. This kind of competition is often quite expensive and indicates that excess returns can be wrest away easily. Therefore the risk of businesses, which are exposed to such changes, is higher than of businesses that sell products with marginal changes. Of course, some companies even maintain their competit ive advantages in fast-changing industries over decades (e. g. Microsoft, Inc. r maybe Facebook) due to customer retention and network e? ects, which create switching costs on the demand side and enormous costs to enter the market on the supply side. The mispricing of such companies that exhibit a durable competitive advantage originates from the complexity in identifying such companies in advance. The following chapter elaborates a bit more on this and points out the state of the art as well as the existing research gap. 16 Chapter 3 Literature Review 3. 1 Competitive Advantage Competitive advantage is a central theme in value investing that has often gone forgotten in the heated debate about the value anomaly.Although an immense number of books and papers have been written about competitive advantage, it has not found proper entrance into the value discussion. Nonetheless, it is an essential part in the valuation process of a company. Greenwald, Kahn, Sonkin, and van Biema (2001) break the Graham and Dodd framework down to three main sources of value (see Figure 3. 1): (1) the asset value, (2) the earning power value, and (3) the value of growth. All three elements must be involved in the calculation of value—also growth (pp. 35-47). The asset value equals the reproduction costs of the assets and is therefore the most reliable source of value.The second most reliable measure of a ? rm’s intrinsic value is the value of its current earnings (earning power value). The earning power value equals current earnings divided by the cost of capital, assuming that the growth rate is zero. The deviation between the asset value and the earning power value equals the franchise of a company. What they call Franchise is referring to the competitive advantage and describes the same phenomenon—the ability to earn more on a ? rm’s assets than it is possible under perfect competition (p. 41). The least reliable source of value is growth, because it i s the most di? ult element of value to estimate and therefore obtains last priority in the valuation process. According to Greenwald, Kahn, Sonkin, and van Biema (2001), growth is only valuable if it is within the franchise. Correspondingly, growth that only increases revenues, earnings or the assets of a ? rm does not create additional value. Growth is valuable only if a company can extend its pro? tability by the means of its competitive advantage. 17 Figure 3. 1: Three Slices of Value Nevertheless, excess returns, which exceed the cost of reproducing a ? rm’s assets, are under the assumption of perfect competition not possible (see Mankiw, 2004, pp. 4-65). As soon as a company earns more on its assets than its reproduction cost, it will attract new competitors, and thus, erode the excess returns until the earning power value equals the value of assets. However that may be, economic theory about perfect competition is seldom the case in reality. Some companies have enjoyed a competitive advantage even over decades (e. g. The Coca-Cola Company or Microsoft, Inc). There have been many research studies conducted on competitive advantage and a huge number of drivers have been found. 1 Without going too deeply into the di? rent drivers, it might be worth to mention the most common: searching costs, switching costs, and economies of scale. By the means of switching costs, a company can create a lock-in: once somebody has chosen a technology, switching can be very expensive (Shapiro 1999, pp. 11-13). Microsoft, Inc. is probably the best example to illustrate a lock-in e? ect. Changing from MS O? ce Word to another writing program is costly. It raises the annoying problem that the formats are not compatible, and thus requires much e? ort that is more costly than remaining with MS O? ce Word. Switching costs can hange over time as buyers alter their products Thomas Fritz (2008) has conducted an extensive literature review of over 140 empirical investigations p ublished between 1951 and 2007. He comes to the conclusion that the di? erent drivers for a competitive advantage are as manifold as the number of studies and that there is no such as a universally valid driver as one could assume. 1 18 and processes (Porter, 1998, p. 296). Another kind of lock-in occurs by search costs. Search costs occur as buyers and sellers attempt to ? nd each other and establish a business relationship (Shapiro, 1999, p. 26). Finally, a competitive advantage arises by economies of scale. Porter (1998) describes economies of scale as the ability to produce more e? ciently at a larger volume (p. 70). But one should note that economies of scale by themselves do not constitute a competitive advantage. In addition to economies of scale, it needs a demand advantage, which does not have to be big. Once a demand advantage exists, economies of scale in the cost structure will transform superior market share into lower costs, higher margins, and higher pro? tability (Gr eenwald, Kahn, Sonkin, and van Biema, 2001, p. 0). Correspondingly, products or services that pro? t from high purchase frequency often enjoy a demand advantage that derives from a habit (e. g. the cigarette industry). Still, it is not written in stone that a competitive advantage lasts for an in? nite period if once achieved. Although a vast number of studies examined the attributes of a ? rm with a competitive advantage, considerably less studies have elaborated on the sustainability of a competitive advantage and the reason why some ? rms enjoy a competitive advantage for decades and other only over a short period. The in? ence of the Competitive Advantage Period (CAP) on the valuation of a ? rm’s shares has also been largely ignored by the literature, although the notion derives its origin from Miller and Modigliani (1961). The term itself appeared in the 90’s in numerous writings. The concept that was developed in Miller and Modigliani (1961)’s seminal pape r on valuation can be summarized as follows: V alue = N OP AT I(ROIC ? W ACC)CAP + W ACC (W ACC) (1 + W ACC) (3. 1) where NOPAT represents net operating pro? t after tax, WACC represents weighted average cost of capital, I represents annualized new investment in working and ? ed capital, ROIC represents rate of return on invested capital, and CAP represents the competitive advantage period. The CAP can be identi? ed, as shown in Equation 3. 1, as a fundamental value driver among risk and cash ? ow. In order to get the CAP we can rearrange Equation 3. 1 as follows: CAP = V alue (W ACC ? N OP AT ) (1 + W ACC) I (ROIC ? W ACC) (3. 2) As Mauboussin and Johnson (1997) assert correctly, this equation has some shortcomings that constrain its practical scope, but it illustrates how the CAP can be con19 ceptualized in the valuation process.According to Mauboussin and Johnson, the key determinants of CAP can be captured by a handful of drivers. The ? rst key determinant is ROIC that re? ects the competitive position within an industry, whereas a high ROIC indicates a strong competitive position. Generally, it is costly for competitors to snatch competitive advantage from high-return companies. The second key determinant is equally important, and measures the rate of industry change. High returns in a fastgrowing industry do not have the same signi? cance as returns created in a stagnated or even shrinking industry. The third driver re? cts the barriers to entry, which is essential for sustainable high returns on invested capital (pp. 68-69). 3. 2 Pro? tability Measurements High-return companies, which have returns in excess of the cost of capital, also capture Warren Bu? ett’s attention. As Mauboussin and Johnson (1997) note, a constant CAP is contrary to economic theory, but it might be achieved through outstanding management. However, companies with a stable CAP are everything but simple to ? nd (p. 71). As mentioned above, Equation 3. 2 has limited practical s cope; thus, in order to evade this problem other performance measures have to be found.In practice, there are many di? erent performance measures, but this thesis will focus in particular on ROE. Fritz (2008) shows in his investigation that ROA and ROE are two of the most frequently applied accounting-based performance measures (p. 31) regarding competitive advantage investigations. Both are pro? tability measurements and capture the relation of return on applied capital. ROE measures how much pro? t a company generates for shareholders while ROA states how e? cient the asset management is. The higher the pro? tability, the better is a ? rm’s economy and the stronger its competitive advantage.Nowadays, less attention is paid to the ROE. Sharpe, Alexander and Bailey (1999) mention the ROE only marginally and Spremann (2007) devote less than one page to it. Nonetheless, ROE has not lost its usability entirely, but Spremann sees the reason for the decreasing importance in the fa ct that shareholders orient themselves more toward market values instead of book values. Provided that, market ratios (e. g. P/E ratio) gained increasingly attention. But since superior earnings are generated based on a competitive advantage, it must remain a core theme in the valuation process, in particular for the long-term investor.Pro? tability measurements tend to change over time; thus, forecasting future profitability is a task that many practitioners and academics would label speculative. On 20 the other side, pro? tability is mean reverting in a competitive environment. Thus, nothing is simpler than predicting long-term pro? tability, which must be zero in the long run. Freeman, Ohlson and Penman (1982) already found evidence that ROE follows a mean-reverting process. Almost twenty years later, Fama and French (2000) found strong evidence of mean-reverting process in terms of pro? ability and estimated a rate of mean reversion of 38% per year. Assuming a ? rm’s ROE of 20% above mean will shrink below one percent after ten years and therefore lose its competitive advantage—,this corresponds to 38% reversion rate. This is also in line with Chan, Karceski and Lakonishok (2003)’s expectation that superior operating performance cannot be sustained for more than ten consecutive years. Furthermore, Fama and French (2000) show that mean reversion is faster below its mean and when it is further from its mean in either direction. However, Penman (1991) scrutinizes ROE regarding its su? iency to predict future pro? tability. According to him, ROE indeed exhibits a mean-reverting tendency, but it proves a too-strong persistence over time. Hence, he suggests that B/M multiples are better indicators of future ROE than current ROE, and a combination of both increases persistence in ROE even further. 3. 3 Research Gap and General Approach Some research has been conducted about predicting future pro? tability. Though these studies deal in particu lar with the issue of predicting the near future. Thus, this study claims high expectations by predicting long-term pro? ability, with the notion that „longterm† means in this study a period of ten years. There are several papers that postulate a mean reversion of pro? tability measures (Freeman, Ohlson and Penman, 1982; Penman, 1991; Lipe and Kormendi, 1994; Fama and French, 2000; Nissim and Penman, 2001). Soliman (2008) forecasts out-of-sample future changes in RNOA ? ve years into the future by applying the DuPont analysis. All these studies have in common that they investigate one ? nancial measure (or two) in time. Thus, this study intends to close these two gaps. In the following chapter, ? rst, several ? ancial measures will be considered regarding companies with a durable competitive advantage, and second, it will be hypothesized that predicting long-term pro? tability (up to ten years) is possible. 21 Chapter 4 Analysis of Long-term Pro? tability The following c hapter aims to determine indicators in order to forecast long-term profitability. Thus, the chapter is structured in four sections: Section 4. 1 describes the data sample and the adjustments. Section 4. 2 deals with the classi? cation of superior performers in terms of ROE and analysis of the persistence of superior performance.Subsequently, the analysis of ROE performance deciles according to persistence is centre stage. Section 4. 3 involves the analysis of further ? nancial measures regarding the ROE persistence deciles. The starting point of this section is the DuPont Identity, which breaks the ROE measure down into further ? nancial measures. The aim of this section is to ? nd speci? c characteristics that will serve in Section 4. 4 to separate ? rms in advance according to future superior performance years. Finally, Section 4. 6 investigates the ROE persistence deciles according to market ratios (i. e. B/M ratio and P/E ratio). . 1 Data Sample A reliable analysis depends to a great extent on the size of the data sample. The size, in turn, is determined by company years (i. e. number of companies times number of years) that are considered. All data in this study originates from COMPUSTAT if there is no explicit mention of it. COMPUSTAT provides historical data of US companies with available historical annual data from 1950. For this study, the dataset on COMPUSTAT was screened for all companies that were listed on any stock exchange in the United States (including inactive companies) with a primary SIC classi? ation between 2000 and 3999. The data was selected at the end of each calendar year between 1979 and 2009. Hence, historical data for the following investigation is available for thirty-one years. Similar to McGahan and Porter (2002), all records from the dataset that do not 22 contain a primary SIC designation after extraction or any that were not within the stated range were dropped out of the sample. The restriction to companies containing a prim ary SIC classi? cation between 2000-3999 corresponds to the manufacturing division, which contains twenty subdivisions (see Table C. ). Focusing on one division has the advantage that the ? rms have a similar value chain. All manufacturing ? rms have in common that they purchase raw materials or components and manufacture these materials to more mature products, which will be sold to a seller or for further processing. Seldom, do these companies sell the product directly to the ? nal consumer. Drawing comparisons among ? rms with similarities regarding their value chain is simpler and also more reliable. Given this restriction to manufacturing companies, 3844 companies are available. It is art of a dynamic industry process that listed companies disappear and new companies appear on trading lists of stock exchanges. This fact leads to certain problems, which were not always considered properly in prior studies. For the sake of convenience, some researchers have considered only compan ies with available data for the entire sample period. Thus, they have excluded companies that were passing through either a delisting or an initial public o? ering (IPO). Others have ignored in their investigation only inactive companies. In this category fall two cases, in particular: Either a company did not survive the entire period due to ? ancial distress and subsequent bankruptcy or it was the target of an acquisition by another company. Ignoring inactive companies would distort the relative ? nancial performance of other companies in the same group in the same period. Not least, since pro? tability depends on competition, it is important to include inactive companies to reduce the e? ect of survivorship bias as it is important to take new competitors into consideration. COMPUSTAT provides the option to also include inactive companies into the sample. Many researchers assume that newly-listed companies show high growth rates that are not economically signi? ant for the compari son to other companies, and thus, lead to distortions (see McGahan and Porter, 2002; Rumelt, 1991; Schmalensee, 1985). Hence, they exclude all companies from the data sample that exhibit less than $10 million in sales. Following these researchers, the sample in this study contains only companies with sales of at least $10 million during the entire sample period. All companies that come below this threshold for any year in the sample period were excluded. After these adjustments, the sample comprises 1905 companies.In order to avoid the possibility that companies distort the calculation of growth rates through short-term measurements, companies with less than ? ve years of ? nancial history were excluded. There is evidence that suggests that window-dressing before an IPO a? ects the performance of subsequent years after the IPO. For instance, Jain and 23 Kini (1994) ? nd that IPO ? rms exhibit a decline in post-issue operating performance (see also Degeorge and Zeckhauser, 1993). The refore, only ? rms with at least ? ve years of ? nancial data on COMPUSTAT items listed in Table 4. 1 were included. Table 4. 1: COMPUSTAT Items This table shows all items hat are downloaded from COMPUSTAT. A more detailed description is given in Appendix A. Companies that have missing data on one of these items are excluded fr