市值老二選股策略 - 政大學術集成

全文

(1)國立政治大學商學院經營管理碩士學程高階經營班碩士論文. 市值老二選股策略 Second is Better. 政治大 A Simple Strategy 立 for Single Stock Selection ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. i Un. v. 指導教授：周冠男博士研究生：張婉珍撰. 中華民國一○六年十二月. 1.

(2) Acknowledgements Investment thinking has been part of my life for decades. Although I have done lots of equity reports and earnings forecasts in my job, this was my first time to do academic research for investment strategy and definitely was a precious and meaningful experience. I would like to express my deepest gratitude to Professor Robin Chou for his valuable and constructive suggestions during the development of this research work. It’s my honor to have Professor Chou as the supervisor and mentor, who always responds to our queries promptly and shares valuable behavioral finance knowledge that inspire me.. 立. 政治大. ‧. ‧ 國. 學. I would also thank my fellow classmates Jason Tsai and Terry Chang, who initiated the team and have always been supportive with great encouragements. Many thanks also go to Fenny Lin, who is attentive to remind me the important timelines, and Dimple Tseng, Bill Chen, Ken, Yu-Jen Lin, all the team members supervised by Professor Chou that have provided the best warmness like a family. Those people who were bouquet givers or audiences in my thesis oral presentation, also deserves thanks for supporting me. In particular I would like to thank Eva Chen and Frank Fu for the Bloomberg data collection in this paper.. n. er. io. sit. y. Nat. al. Ch. i Un. v. I am very grateful for sharing the moments with these people for all the meals, discussion, and laughs. I enjoy this journey so much.. engchi. Wanchen Chang December 2017. 2.

(3) 摘要大型股過去一直被認為平均報酬率低於小型股，但如果從個股來看，不少大型股的績效並不會比指數差。考慮到一般非專業投資人在投資股票時，選擇大型股還是比小型股容易，本論文試圖建構一套在實務上較可行的大型個股選股策略—選擇市值第二大的股票，並定期調整個股。我們以美股標準普爾 500 指數中前兩大市值的股票，分為兩種投資組合做比較，結果發現，市值最大的股票不容易創造超額報酬，市值第二大的股票，反而締造極佳的超額報酬，此現象在過去 3 年、5 年、10 年，尤其較過去 20 年更為明顯。原因在於市值排名第二的股票，多半屬於排名仍在持續上升的成長股，這些個股基本面尚未到達頂點，故股價還會反應一段時間的基本面利多，採取類似動能策略(Momentum Strategy)的方法，報酬率容易超越指數；市值最大者則因為基本面普遍伴隨市值排名已經到頂，加上投資人對於排名第一的股票，多半易產生定錨效應(Anchoring Effect)，即認為股價可能已經反應其該有的價值，較難創造超額報酬，傾向賣出。故同樣投資大型股，選擇市值第二名的股票會優於第一名。. 立. 政治大. ‧ 國. 學. 關鍵字：大型股、市值排名、超額報酬、動能策略、定錨效應、股價. ‧. n. er. io. sit. y. Nat. al. Ch. engchi. i. 3. i Un. v.

(4) Abstract According to The Size Effect Theory, small cap securities generally generate greater returns than those of large cap companies. However, this trend has involved into the difficulties of stock picking due to the large number of small caps. In this paper I propose a strategy against the size effect theory, “Second is Better”, to pick the second largest market value security as the single stock investment. I examine the performances of the No.1 and the No.2 largest market cap stocks in the S&P500 and apply a 6-month rebalance to construct two different portfolios, which is similar to the concept of. 政治大 stock outperforms than No.1 stock and generate amazing excess returns in the near 立 Momentum Strategy that buy the past winners and sell the past losers. I find the No.2. ‧ 國. 學. mid-to-long-term periods. Because No.1 stocks are more likely to experience Momentum Crash than No.2 stocks due to investor’s anchoring bias as they believe the No.1 stock. ‧. might have been peaked. No.2 stocks are usually in the growing stages that many. al. er. io. sit. y. Nat. investors believe the 2nd largest caps still yet to peak during market value expansion.. v. n. Keywords : Size Effect, Market Value, Large Cap, Momentum Strategy, Anchoring Bias. Ch. engchi. ii. 4. i Un.

(5) Content 1. Introduction .......................................................................................................................1 1.1. Motivation of the study ..............................................................................................1. 1.2. Chapter outlines .........................................................................................................3. 2. Literature review ...............................................................................................................4 2.1. Size Effect ....................................................................................................................4. 2.2. Concentration vs. Diversification................................................................................6. 2.3. Momentum Strategy and Anchoring Effect ................................................................7. 政治大 3. Single Stock Selection Strategy ........................................................................................11 立 Earning Growth Comparison of Large and Small Caps ...............................................8. 3.1. Interpretation ...........................................................................................................11. 3.2. Data Collection and Methodology............................................................................11. ‧ 國. 學. 2.4. ‧. 4. Empirical Results..............................................................................................................14 Returns of stock selection strategy ..........................................................................14. 4.2. Explanation of the Divergence of the Two Portfolios ...............................................17. er. io. sit. y. Nat. 4.1. al. iv n C hengchi U 5. Conclusion .......................................................................................................................25. Why Does Portfolio B Outperform? .........................................................................22. n. 4.3. 6. References .......................................................................................................................28 7. Appendix ..........................................................................................................................30. iii. 5.

(6) List of Tables Table 2-1: PER, Index Gains, Accumulated EPS Growth of S&P100 and Russell 2000 ·····10 Table 4-1: Risk-adjusted Returns of Portfolio A, B, C ·······················································16 Table 4-2: Stocks’ 6-mth Returns of Portfolio A ·······························································20 Table 4-3: Stocks’ 6-mth Returns of Portfolio B ·······························································21 Table 4-4: Holding stocks in Portfolio B by ranking upgrade/downgrade ························22. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. iv. 6. i Un. v.

(7) List of Figures Figure 7-1 Ranking change of GE......................................................................................30 Figure 7-2 Op margin of GE ..............................................................................................30 Figure 7-3 Ranking changes of KO ....................................................................................30 Figure 7-4 Ranking changes of XOM ................................................................................31 Figure 7-5 Op Margin of XOM ..........................................................................................31 Figure 7-6 Ranking changes of MSFT ...............................................................................31. 政治大 Figure 7-8 Ranking changes of INTC .................................................................................32 立. Figure 7-7 Op margin of MSFT .........................................................................................32. ‧ 國. 學. Figure 7-9 Ranking changes of AAPL ................................................................................32 Figure 7-10 Op Margin of AAPL ........................................................................................33. ‧. Figure 7-11 Ranking changes of GOOGL ..........................................................................33. n. er. io. sit. y. Nat. al. Ch. engchi. v. 7. i Un. v.

(8) 1. Introduction 1.1 Motivation of the study Many individual investors are facing difficulties to look for single stock investments. They might do stock selection either based on fundamental views or technical ways. But think how much time and efforts it might take before come out the best ideas. Is there any easy way to pick a good company without lots jobs? Many “good” stocks have been discovered by the global investors to be part of their portfolios. The stocks with larger market caps usually mean they are more favorable than. 政治大. others in the market, given more investors are holding them. It would be ideally considered. 立. those top large caps as “good” stocks in a simple way without doing any research, since those. ‧ 國. 學. global institutional investors have proved it through their actions.. ‧. But “good” large caps may not equal to those “outperforming” stocks. Small and midcaps ruled the four-year period between 2011 and 2014. However, increased regulations. y. Nat. er. io. sit. and earnings weaknesses among smaller companies all point to large-cap leadership in 2016. We can say there is a decided advantage for large caps in terms of liquidity and research. n. al. Ch. i Un. v. coverage as they tend to operate with more market efficiency than smaller segments. Large. engchi. caps might outperform than others more easily if most investors are buying this idea. However, this idea is contradictory to the Size Effect Theory, which stresses the fact that small-cap stocks tend to yield higher risk-adjusted returns. There is a widely held belief that small caps tend to outperform than larges caps. But it is too difficult for individual investors to do stock picking due to the large number of small cap stocks. They might buy a basket of small caps to diversify the risks. Though diversification has been viewed as risk-reducer to successful investing, it is still inefficient for individual investors to do so because of too many stocks to track. I believe that it should be easier and efficient to select the single stock from large caps 1. 8.

(9) given the limited choices than small caps. The information is also more readily available for retail individual investors because of the wide coverage of press media due to the importance of large caps which are usually the leading industry players. They are hardly to be neglected by all the investors or reporters. Although the neglect effect is exactly how the researchers explain the outperformance of small caps due to information inefficiency, I don’t think it would be pros for individual investors to buy small caps given the poor exposures cause very high risks. For example, delays of new iphone product launch of Apple would be big news that there will be plenty of stories or comments hardly to be missed. But you may not notice a small company with the same issue because it is not the focus of media.. 政治大. This belief constructs my thoughts to find a simple way to do single stock selection from. 立. large caps after seeing a few of super large market value stocks did outperform than index. ‧ 國. 學. fund recently. Can those top 1 or 2 market cap stocks generate excess return if we only pick. ‧. them as our portfolio?. In this paper, I am going to introduce a strategy that considers the second largest market. y. Nat. er. io. sit. value firm as the better single stock for individual investors. The strategy, which I am going to call “Second is Better”, works by rebalancing on a 6-month basis to invest only in the second. n. al. Ch. i Un. v. largest market value stocks. My study proves that the very large market value stocks can. engchi. outperform than the small caps in some cases that are easily to be identified for individual investors. The reason why “Second is Better” attributes to the fact that most No.2 stocks are in the process of market value ranking upgrading, implying they are still in growing, that the investors believe their fundamentals are yet to peak comparing with the No.1. The concept is similar to the traditional momentum strategies that long the top 10% of past 12-month winners and short the bottom 10% of past 12-month losers generate positive returns over time. But my strategy is much more concentrated on the single stock in order to identify the differences between buying the No.1 and No.2 stock. If both of them work under the same approach, it means the best strategy for individual investors is to buy the top 2 market value 2. 9.

(10) companies. However, the results of my study only support No.2 stock does outperform in most of the cases. The interesting difference could be also explained by the Investors’ anchoring effect that people might lose interests in the No.1 stock as they believe the stock might have been peaked.. 1.2 Chapter outlines The piece consists of three parts: In the first part (Chapter 2), I am going to review the literature of the Size Effect Theory. 政治大 diversification. The literatures about Momentum Strategy will also be reviewed as well as 立. and the debates about it. I will also discuss the studies that against the traditional idea of. ‧ 國. 學. how the researchers explain Momentum Crash, which is often seen in the Momentum Strategy that would be good references to examine the failures of No.1 stock in my portfolio.. ‧. In Chapter 3, I will propose my single stock selection strategy “Second is Better” as the. sit. y. Nat. core research of this paper. There are two portfolios constructed according to the rankings of. io. er. market values, one is the largest while another is the second largest cap.. al. iv n C h esecond Chapter 4. I’m going to discuss why the market value stock outperforms whilst hi U n g clargest n. Their performances will be examined and followed by an explanation of divergence in. No.1 stock fails in my study. The paper concludes with a brief summary.. 3. 10.

(11) 2. Literature review 2.1 Size Effect There are a large number of studies discussing size and its effect on risk and return during the past years. Banz (1981) was one of the first researchers who noticed an evident negative relationship between the market values and the yielding return of the firm’s stocks. The fact that small-cap stocks tend to yield higher risk-adjusted returns is known as the size effect. The idea of earning premiums for investing in smaller companies has become an important consideration for investors and led to the introduction and widespread use of small-cap funds.. 立. 政治大. Fama and French (1992, 1995) also find that size and P/B independently explain returns.. ‧ 國. 學. The Fama and French Three Factor Model expands on the capital asset pricing model (CAPM). ‧. by adding size and value factors to the market risk factor in CAPM. This model considers the fact that value and small-cap stocks outperform markets on a regular basis.. y. Nat. er. io. sit. However, when examining the traditional small-minus-big value adjusted long/short factor (SMB) developed by Fama and French, some researchers found the performance over. n. al. Ch. i Un. v. the past 30 years (1983-2013) has been flat and highly volatile. Wesley R. Grey (2014). engchi. created an extreme size value-adjusted factor (E-SMB) which is based on quintile splits of size rather than median splits used by the original SMB factor. If the size effect does work, it should be larger premium associated with E-SMB. But the result shows E-SMB has WORSE performance relative to SMB. The author concludes the size effect exists only in long-only portfolios, which are equally weighted according to Banz’s approach, rather than the long/short portfolios conducted from Fama-French method. Some economists do not agree that the small-firm effect really exists. There are several different explanations for the underlying effect can be found, one of which is the neglected-firm effect. According to the hypothesis, firms neglected by analysts or investors 4. 11.

(12) should obtain substantially returns to compensate for the gap of information accessibility. Dowen and Bauman (1986) find that size fully explains the neglect effect. Small firms usually suffer from lack of public information and therefore are neglected by professional investors. As such, investors may require additional returns for holding small-neglected stocks. There is another argument that small stock premium is derived from a single observation as much of the historical premium was earned in a single period from 1974 to 1983. Jeremy Siegel (1994) indicated that the total accumulation of small stocks falls nearly 25% below that of large stocks if the period from 1974 to 1983 is eliminated. The key reason that causes small cap stocks to experience a strong upward move during 1970s and early. 政治大. 1980s is believed to be the “neglect effect”. The neglected companies were undervalued at. 立. the start of the period, which were trading at P/E ratio 50% lower than that of S&P500. After. ‧ 國. 學. the big investment houses became interested in these stocks and made a massive purchase,. ‧. the share prices of small cap stocks were driven up. By 1983, small cap stocks were trading at 20% P/E ratio premium compared to that of S&P500.. y. Nat. er. io. sit. Dijk (2005) observed there is lots of variation in the size effect over time that the size premium is negative in 38 out of 84 years (1927-2010). He indicated “averaged over the past. n. al. Ch. i Un. v. 30 years, the return difference is only 1.2% per year. Especially in the periods 1946–1957 and. engchi. 1980–1999, small stocks clearly underperformed large stocks”. Investor behavior could be also one of the explanations to see why small stocks underperform. Gompers and Metrick (2001) suggest that the growing US equity market held by institutional investors has enhanced the demand for large and liquid stocks. They found that large institutional investors almost doubled their share of the stock market from 1980 to 1996. This compositional shift tends to increase demand for large companies and decrease demand for the small caps, which can partially explain the disappearance of small stock’s premium. The latest study also suggests the tides of size effect have shifted recently due to 5. 12.

(13) investor behaviors that support a bias toward large caps over small caps. Liz Ann Sonders (2017) observes the cumulative advance/decline lines for the S&P 500 (a large cap index) and Russell 2000 from Mar/2016 to Mar/2017 have diverged more recently in favor of larger caps. This could be explained by the fund flows toward the larger cap passive exchange-traded funds (ETFs). Moreover, small caps are trading at a 10% premium to large caps on 2017 earnings and 6% on 2018 earnings. This implies small caps might not be undervalued as it was before.. 2.2. Concentration vs. Diversification. 政治大. Since my study focuses on the single stock selection strategy, I am going to review the. 立. literature discussing about diversification and concentration for investment in this section.. ‧ 國. 學. The Modern Portfolio Theory (MPT) highly emphasizes the importance of diversification. ‧. to reduce risks. But many famous investors such as Warren Buffett, George Soros and Bill Ackman have proved to be successful through more concentrated portfolios. Yeung et al.. y. Nat. er. io. sit. (2012) indicates there is a trade-off between diversification and returns. He finds that the excess returns increased accompany with a more concentrated portfolios. His findings. n. al. Ch. i Un. v. suggest that fund managers have good stock-selection skills as their top ideas tend to. engchi. outperform. He concludes that performance suffers when portfolio managers attempt to diversify holdings by moving to those stocks which are mot their top ideas. Cohen, Polk, and Silli (2010) also conducted a study to evaluate the performance of the best ideas in US equity mutual funds. They found that portfolio’s best ideas not only generated significant risk-adjusted returns over time but also outperform the rest of the positions. Higher risks might be the key consideration for the concentrated portfolios in many investors’ eyes. However, Elton and Gruber’s book (2009) indicates that adding too many securities for diversification purposes only leads to marginal risk-reduction results. According 6. 13.

(14) to the book, the additional stocks between 20 and 1,000 only reduced the portfolio’s risk (standard deviation) by 0.8% percentage points. In a word, the literature above could support my study in digging the large market value stock as the single stock portfolio, which is against the size effect theory that small stocks tend to outperform, and the Modern Portfolio Theory that diversification is the key to successful investing.. 2.3 Momentum Strategy and Anchoring Effect My strategy that buying the top market value stocks is consistent with the conventional. 政治大. momentum strategy by buying past winners and selling past losers. Jegadeesh and Titman. 立. (1993) are the first to document the momentum strategy by sorting the firms based on the. ‧ 國. 學. 3-to-12-month past returns in U.S stock market. Their result shows that buying the top 10%. ‧. and selling the bottom 10% of stocks ranked by returns during the past 6 months, and holds the positions for 6 months, generates profits of 1% per month. There is no short strategy in. y. Nat. er. io. sit. my approach as it is difficult for individual investors to find the bottom small caps to short in reality. But the ideas that past winners tend to produce positive returns are similar. The top. n. al. Ch. i Un. v. large caps usually represent the winners in the market as their share price total returns have. engchi. been outperforming than peers for a very long time to become the No.1 or No.2 large cap stocks. However, despite the strong performance of momentum strategy, Daniel and Moskowitz (2014) indicate it can experience infrequent and persistent negative returns during panic state, which is called “momentum crashes”. The occurrences of momentum crashes are concentrated on the market rebound. Suk Joon Byun and Byoung Hyun Jeon (2015) suggest crashes are because of investor’s anchoring bias. As the market dramatically rebounds from the downturn, investors seek stocks that will rebound the most. Stocks that are far from their previous price peaks would be the names that believe to have higher room 7. 14.

(15) to run up. Thus stocks with high anchoring price will outperform those with low anchoring price. This supports my study that No.1 market cap stocks underperform than the index average due to investor feel the No.1 stock has lower room to run.. 2.4 Earning Growth Comparison of Large and Small Caps I investigate the fundamental changes in different time periods of large caps and small caps in this section in order to find the explanations of size effect deterioration. Data are constituted from the S&P100 and Russell 2000 index values in a monthly basis over the period 31/Jan/1990-2017/5/30 and are sourced from the Bloomberg database. I selected. 政治大. S&P 100 index and Russell 2000 index values to represent large caps and small caps,. 立. respectively.. ‧ 國. 學. The S&P 100 index, a subset of the S&P 500, includes 102 leading U.S. stocks with. ‧. exchange-listed options. Constituents of the S&P 100 are selected for sector balance and represent about 63% of the market capitalization of the S&P 500 and almost 51% of the. y. Nat. er. io. sit. market capitalization of the U.S. equity markets as of January 2017. The stocks in the S&P 100 tend to be the largest and most established companies in the S&P 500.. n. al. Ch. i Un. v. The Russell 2000 index is an index measuring the performance approximately 2,000. engchi. small-cap companies in the Russell 3000 Index, which is made up of 3,000 of the biggest U.S. stocks. The Russell 2000 serves as a benchmark for small-cap stocks in the United States. I compare the Price to Earing Ratio (PER), average EPS growth and the index gains of S&P 100 and Russell 2000 during Year 2001~2016 and Year 2006~2016 as presented on Table 2-3. Theoretically, if the accumulated index gains are much higher than the earnings growth, it means the stocks might have been fairly valued or overvalued. In other words, the index stocks are undervalued if the index gains are lower than EPS growth. The total accumulated index gain of Russell 2000 during 2001-2016 is 177.8%, which approximates its accumulated EPS growth (188.5%) at the same time period. However, the 8. 15.

(16) accumulated EPS growth of S&P100 during the same period is 132.5%, much higher than the index gains of 69.7%. This implies that S&P100 is undervalued while Russell 2000 is fairly valued. On valuation wise, Russell 2000 index has been trading in a much higher PER than S&P 100. The average PER of Russell 2000 during period Year 2001 to 2016 is 28.1x, while that of S&P100 is 15.9x. The relatively higher PER of small caps is supposed to be adjusted by stronger earnings growth compared to large caps. However, when looking into the numbers during Year 2006-2016, we find Russell 2000 only accumulates total EPS growth at 10.3%, which is half of that of S&P 100. Although the total accumulated index gain of Russell 2000 is 72.3%, still much higher than that of S&P 100. 政治大. (50.1%), investors’ incentives to long small caps may diminish as small caps are no longer to. 立. produce higher profit growth momentum than large caps as they were before.. ‧ 國. 學. We can say size effect may exist in many of the cases, but not that much as time goes by.. ‧. Recap Banz’s approach to analyze the size effect, we find his theoretical portfolios are designed to be equally weighted, held for 5 years and rebalanced every month during the. y. Nat. er. io. sit. 1936-1975 time period. It is different from the index portfolios that are value weighted and daily rebalanced. It is hard to replicate the same approach as Banz nowadays to the real. n. al. Ch. i Un. v. market implementation because of the trading costs for monthly rebalance and difficulties to. engchi. define which stocks are the small caps to buy. Individual investors are always easier to identify those large market value stocks rather than small ones from the rankings. The above academic research and study all reinforce my strategy thinking in the next chapter—to find an easy and best way to single stock selection from the companies with top market values.. 9. 16.

(17) Table 2-1: PER, Index Gains, Accumulated EPS Growth of S&P100 and Russell 2000 This table compares the accumulated index gains and earing growth between S&P100 (Large Caps) Russell 2000 (Small Caps) during two different time periods, Year 2001~2016 and Year 2006~2016. Although Russell 2000 reports stronger earnings growth and higher accumulated index returns during 2001-2016, S&P100 (Large Caps)’s index gains are only half of its earnings growth, implying large caps might have been undervalued in the past 15 years. Small caps (Russell 2000) then reports weaker total EPS growth but higher accumulated index gains than large caps during 2006~2016, implying it might be overvalued in decade.. Base 2001= 100 Accumulated Index gains. PER Russell 2000. S&P 100. Accumulated Index ave. Accumulated Index Accumulated Index EPS growth gains ave. EPS growth. Russell 2000. S&P 100 Russell 2000 S&P 100. 2001. 23.9. 27.9. 2002. 19.2. 42.3. -23.9%. -21.6%. 6.0%. 2003. 17.7. 46.9. -5.7%. 14.0%. 31.9%. -6.4%. 2004. 17. 32.3. -1.5%. 33.4%. 53.6%. 138.5%. 2005. 15.82. 23.7. -2.4%. 37.8%. 63.3%. 117.9%. 2006. 15. 22.6. 13.0%. 61.2%. 93.4%. 161.5%. 2007. 15.3. 24.1. 17.3%. 56.8%. 92.8%. 142.3%. 2008. 13.6. 22.4. -26.1%. 2.2%. 44.0%. 2009. 15.2. 37.5. -12.0%. 28.0%. 38.0%. 12.7. 20.5. 2013. 14.2. 24.2. 2014. 15.6. 26.4. 2015. 16.8. 27. -2.7%. -0.3%. -7.4%. -47.4%. -34.7%. -36.6%. -25.5%. -79.9%. -32.1%. -22.2%. -20.6%. -28.7%. -74.0%. -3.1%. 60.4%. 89.8%. 111.5%. -14.3%. -0.5%. -1.9%. -19.1%. -2.3%. 51.7%. 123.5%. 143.6%. -13.6%. -5.9%. 15.6%. -6.9%. 10.7%. 73.9%. 126.5%. 174.4%. -2.1%. 7.8%. 17.1%. 4.9%. 41.0%. 138.2%. 137.3%. 150.0%. 24.7%. 47.7%. 22.7%. -4.4%. 55.5%. 146.6%. 148.8%. 188.5%. 37.5%. 56.0%. 132.5%. 130.7%. 200.0%. 177.8%. 132.5%. 188.5%. 26.5 28.1. n. 17.3 15.92. io. 2016 Average. al 69.7%. Ch. engchi. 10. 17. 52.9%. 28.7%. 10.3%. 38.0%. 44.2%. 19.3%. 14.7%. 50.1%. 72.3%. 20.2%. 10.3%. er. Nat. 23.4. 21.9. y. 2012. Russell 2000. 3.8%. ‧. 13.2 12.2. S&P 100. -20.5%. 學. 2010 2011. ‧ 國. 立. Russell 2000. 政治大. sit. S&P 100. Base 2006= 100. i Un. v.

(18) 3. Single Stock Selection Strategy 3.1 Interpretation The above study shows that the conventional theories such as size effect and diversification could be a myth. Even if the small caps might yield higher returns than large caps according to the result of statistics, it doesn’t mean it is good to invest our money into those small caps which we barely know their fundamental details and risks, regardless to say which one would be considered to be our best single stock target. In my view, it is meaningless to compare the average return between large caps and small caps before. 政治大. making an investment decision given we still don’t know which stock would be the best idea. 立. after testing the size effect. The strategic thinking should be focused on how to find a good. ‧ 國. 學. stock which is more easily to beat the index fund.. ‧. The below study emerged from something interesting I found during the process of examining the size effect. The individual stocks with the leading market capitals seem have. y. Nat. er. io. sit. excess returns in some cases, if comparing with the index fund. By only picking those top market caps as our single stock targets and applying periodic rebalance to eliminate those in. n. al. Ch. i Un. v. the downturn trend, could it be a useful tool for us in equity investment?. 3.2. engchi. Data Collection and Methodology. This is a single stock selection strategy idea which only invests in the leading market cap firm to see whether it outperforms than the index fund or not. I decide not to include too many large cap stocks but only the top 2 to keep it clean and simple. I choose the S&P 500 as my database. The reason I sourcing the market data from the United States rather than from Taiwan or other Asian countries is because 1) the largest market cap stock trading in the Taiwan, Hong Kong or Japanese Stock Market has been always occupied by single company for a very long time, which is difficult to observe the 11. 18.

(19) rotation effect, 2) the stock with the top market cap in the S&P500 has changed around 10 times in the past 20 years, which is good to examine my strategy. I also believe the U.S. stock market has the highest efficiency well to do sampling. The share prices and market value data was sourced from the Bloomberg that spanned from 1996/12/30 to 2017/6/30. The initial date was targeted at end-1996 given a more precise period in calculating 20-year total returns. Monthly total stock returns are calculated by the share prices of the last trading day on each month. All stocks in S&P500 are ranked by market values in descending order on every rebalance day in order to find which one has the largest market caps. 9 companies are. 政治大. selected into the pool given 4 of them were ranked as the top market-value stocks during. 立. different time whilst the rest were the 2nd largest market caps in the S&P 500 during different. ‧ 國. 學. periods.. ‧. The initial fund is assumed to be $1m and fully invested without considering the extra expenses of trading costs or tax. The rebalance date is set to be end-Jun and end-Dec. The. y. Nat. er. io. sit. frequencies of 6-month rebalancing are based on the reasons that, firstly, the top 2 companies ranked by market values only changes once a while within one or two years, or. n. al. Ch. i Un. v. even longer because of the super huge market values are hardly to be outpaced by other. engchi. stocks. Hence, 6-month rebalances should be long enough to reflect in-time ranking changes. Secondly, most of the momentum portfolios are formed based on past 6-month returns, and hold the position for 6 months, as we can find from the research conducted by Jegadeesh and Titman (1993) and Moskowitz and Grinblatt (1999). It would be ideally to adopt the similar approach since my strategy is elaborated from the concept of momentum strategy. There are 3 kinds of single stock portfolio. Portfolio A is always to hold the stock that has the largest market values. Portfolio B is to own the second largest market cap stock. Portfolio C is to equal weight the holdings of Portfolio A and B. The methodology underlying the portfolio construction is described as below: 12. 19.

(20) Portfolio A: In the beginning day (30/Dec/1996) of investment, I buy the stock X1 that has the top market values in the S&P500 and hold it until the next rebalance day, which would be 30/Jun/1997, six month later. If X1 remains the top cap stock by the rebalance day, then no action will be taken. However, if X1 is replaced by another company, let’s say X2, I will do stock switch to X2 from X1 on the rebalance day. The selling/buying prices of X1 and X2 are defined to be their last prices of the rebalance day for the purpose of quick and dirty calculation. Portfolio B: The rebalance mechanism is the same as Portfolio A but the target stock Y1 is the one with the second largest market cap.. 政治大. Portfolio C: The fund is split into 50:50 that invests in both Portfolio A and Portfolio B. 立. with equal weight.. ‧ 國. 學. The total return, annualized return and the standard deviation of Portfolio A, B, C will. ‧. then be compared with that of S&P 500 index fund, S&P 100 index and Russell 2000 index during four kinds of time interval: 1997~2017, 2007~2017, 2012~2017, and 2015~2017. This. y. Nat. understand my single stock selection strategy works or not.. n. al. Ch. engchi. 13. 20. er. io. sit. is for the purpose of comparing what might be changed in different periods. We then can. i Un. v.

(21) 4. Empirical Results 4.1 Returns of stock selection strategy In this section, I evaluate the performance of my single stock selection strategy and compare it with the indexes. Table 4.1 shows the average annualized returns, standard deviation and Sharpe Ratio during four different holding periods for Portfolio A, B, C versus S&P 500 index fund, S&P 100 index, and Russell 2000 index. Portfolio A doesn’t prove to be successful as the 20-years, 10-years and 5-year annualized returns are all below those of S&P 500 index fund by 1.03%, 1.12% and 6.67%,. 政治大. respectively. Only the investment during 2015~2017 generates a significant excess return of. 立. 4.52% (Table 4-1).. ‧ 國. 學. Surprisingly, Portfolio B shows a contrary result that the excess returns are all positive in. ‧. different periods. The average annualized excess return increased considerably over time, from 0.85% over the 20-year period, to 12.43% over the period 2015~2017. It is also worth. y. Nat. er. io. sit. noting the numbers during 2012~2017 comparing with Portfolio A, which has the worst performance of negative excess return of -6.67%, whilst Portfolio B reaches a very good. n. al. excess return of 9.77%.. Ch. engchi. i Un. v. As for the Portfolio C, the excess returns are in between of Portfolio A and B as expected since it is an equally weighted portfolio. This balance strategy offsets the poor performance of Portfolio A and delivers positive 10-years, 5-year and 2.5-year excess returns of 1.56%, 1.12%, and 8.95%, respectively. Only the 20-year performance is almost the same as the benchmark without excess return. The comparison of Sharpe Ratio, which is used to measure the risk-adjusted return, also shows an interesting result that Portfolio B is ranked top 1 both in the period 2007-2107 and period 2015-2017. Not surprisingly, Portfolio A has the worst Sharpe Ratio during mid-to-long-term period including Year 1997-2017, Year 2007-2017, and Year 2012-2017. It is 14. 21.

(22) known that single stock investment usually poses high risks with high standard deviation that may damper Sharpe Ratio. But the strong returns of Portfolio B compensate the risks. It is noted that the Sharpe Ratio of Portfolio B is only 0.55, lower than other indexes during the longest observing period, Year 1997-2017, though it has the second highest annualized return. S&P 500 index reports the highest Sharpe Ratio of 0.83, followed by Russell 2000, which records Sharpe Ratio at 0.78, during the same period. Russell 2000’s Sharpe Ratio has been deteriorated and records as the lowest during Year 2015-2017. S&P100 doesn’t produce better Sharpe Ratio than S&P500, either. The result shows both large caps index (S&P100) and small caps index (Russell 2000) report weaker. 政治大. Sharpe Ratio than the benchmark(S&P 500). The table also illustrates that S&P 500 performs. 立. quite well in terms of risk-adjusted returns, which reports the best Sharpe Ratio in the past. ‧ 國. 學. 20 years and period Year 2012-2017. The average risk-adjusted returns of S&P 500 are only. ‧. behind that of Portfolio B during Year 2007-2017 and Year 2015-2017 if excluding Portfolio C. The indexes’ annualized returns in different periods also look interesting. If we only look. y. Nat. er. io. sit. at the 20-year annualized return of the three indexes in Table 4.1¸ the Size Effect seems existed as Russell 2000 index reports the highest returns in all comps. However, S&P 500. n. al. Ch. i Un. v. index fund has the best risk-adjusted returns (Sharpe Ratio), which doesn’t support the Size. engchi. Effect Theory. The more recent data, especially those after Year 2012, all point to the fact that small caps index (Russell 2000) has weaker risk-adjusted returns than that of S&P 500 and S&P 100.. 15. 22.

(23) Table 4-1: Risk-adjusted Returns of Portfolio A, B, C This table presents the average annualized returns, standard deviation and Sharpe Ratio during four different holding periods for Portfolio A, B, C versus S&P 500 index fund, S&P 100 index, and Russell 2000 index. The Sharpe Ratios represent the risk-adjusted returns of all portfolios, which are calculated as = (Mean portfolio return − Risk-free rate)/Portfolio Standard deviation. The U.S. 10-year treasury yield, 2.38%, is defined as Risk-free rate.. Benchmark Large Caps Small Caps Portfolio A Portfolio B Portfolio C S&P 500 Index Fund. 288.71 6.85 0.89 5.7 0.78 1 2. 167.93 4.93 -1.03 7.55 0.34 6 6. 285.9 6.81 0.85 8.06 0.55 2 5. 226.92 5.95 -0.01 6.41 0.56 4 4. 70.74 5.23 NA 4.27 0.67 4 3. 61.67 4.68 -0.55 4.16 0.55 5 5. 76.69 5.74 0.51 5.63 0.60 3 4. 52.61 4.11 -1.12 5.91 0.29 6 6. 152.96 9.24 4.01 6.98 0.98 1 1. 99.27 6.79 1.56 4.94 0.89 2 2. 92.7 12.7 NA 2.86 3.61 3 1. 87.1 12.1 -0.6 2.87 3.39 5 2. 91 12.5 -0.2 3.98 2.54 4 4. i Un. 37.97 6.03 -6.67 6.59 0.55 6 6. 204.87 22.47 9.77 6.06 3.32 1 3. 103.83 13.82 1.12 4.95 2.31 2 5. 17.54 6.68 -0.03 3.16 1.36 5 4. 17.49 6.66 -0.05 4.18 1.02 6 6. 30.48 11.23 4.52 6.67 1.33 3 5. 54.93 19.14 12.43 6.43 2.61 1 1. 43.88 15.66 8.95 5.34 2.49 2 2. 政治大. n. al. Ch. 17.64 6.71 NA 3.08 1.41 4 3. engchi. 16. 23. sit. er. io. 196.59 5.45 -0.51 4.42 0.69 5 3. y. Nat. 227.45 5.96 NA 4.32 0.83 3 1. ‧. ‧ 國. 立. Equal Weight Russell No.1 No.2 on Portfolio 2000 Index Largest cap Largest cap A and B. 學. 1997-2017 Total Return Annualized Return Excess Return Standard Deviation Sharpe Ratio Ranked by Annualized Return Ranked by Sharpe Ratio 2007-2017 Total Return Annualized Return Excess Return Standard Deviation Sharpe Ratio Ranked by Annualized Return Ranked by Sharpe Ratio 2012-2017 Total Return Annualized Return Excess Return Standard Deviation Sharpe Ratio Ranked by Annualized Return Ranked by Sharpe Ratio 2015-2017 Total Return Annualized Return Excess Return Standard Deviation Sharpe Ratio Ranked by Annualized Return Ranked by Sharpe Ratio. S&P 100 Index. v.

(24) 4.2 Explanation of the Divergence of the Two Portfolios In this section, I am going to discuss why Portfolio A and B, the No.1 and No.2 market value stocks, perform totally different results in my investigation. The key explanation would be whether the stocks are upgraded or downgraded in terms of market value ranking. Portfolio B has more upgraded stocks. However, Portfolio A has no upgraded but only “downgraded” stocks, given they are already ranked as the top 1 cap. Portfolio A is largely underperformed because the share prices of MSFT slumped by. 政治大. over than 30% during the first half year of 2000 (Table 4-2). The holding stock then switched. 立. to GE by 30/Jun/2000 which just replaced MSFT as the largest cap firm. However, GE’s share. ‧ 國. 學. price also saw a huge drop by more than 40% from Jun/2000 to Jun/2002 even though it is still the stock with largest market values. The rebalance on Jun/2002 then did another switch. ‧. to MSFT from GE as MSFT was back to top. But unfortunately, MSFT didn’t work in the. Nat. sit. y. following 12 month that the share prices declined by more than 20% during. n. al. er. io. Jun/2002~Jun/2003. MSFT lost its market cap crown therefore the rebalance brought GE. i Un. v. back into the portfolio again on Jun/2003. Nevertheless, GE didn’t create much excess return. Ch. engchi. during 2003~2005. Portfolio A finally saw excess return after switched to XOM by Jun/2006 but still suffered from collapse in financial crisis. The rebalance switches between XOM and AAPL during 2012~2103 is another bad strategy to draw down the performances. Buying AAPL on Dec/2013 is a turnaround of performance with excess total return of 26.4% in 2014, but not good enough to offset the impacts from the previous trades when looking at the annualized total return including 20 years, 10 years and 5 years. Portfolio B, on the contrary, enjoys several sweet points that contribute lots to performances (Table 4-3). The switch action that buys MSFT and sells GE by end-2008 is so successful that generates excess returns by 33.3% in 2009. This strategy also achieves 17. 24.

(25) amazing excess returns of 76% in 3 years from 2011 to 2013 by holding AAPL before Jun/2012 and switching it to XOM after Jun/2012. Buying GOOGL on Jun/2015 is another key contributor that generates excess return by 30%. There were a few of periods that Portfolio B delivers relatively poor performance such as Year 2000 (total return -53.79%, negative excess return -43.11%) due to switch to INTC by Jun/2000, which was plunged by 55% in the next 6 month. It is underperformed during 2007~2008 that attributed to the holdings of GE, which is downgraded from Portfolio A by end-Jun 2006. The performance in 2014 is also very weak with negative excess return of -19.94% because of the share holdings of XOM, which was kicked out from Portfolio A by end of 2013.. 立. 政治大. We found the loser stocks in Portfolio B have something in common. Most of them are. ‧ 國. 學. downgraded from Portfolio A, which means their share price movements are down trending. ‧. from the peak. On the contrary, the winners that attribute remarkable excess returns to Portfolio B are usually those upgraded from a lower rank, implying a positive up trend of. y. Nat. er. io. sit. share prices. These “winner” stocks perform so well that offset the impacts of the “losers” on a mid-to-long-term basis. The concept is like Momentum strategy, which is typically. n. al. Ch. i Un. implemented by buying past winners and selling past losers.. engchi. v. I classify the total 7 names that have been selected into Portfolio B in different periods by ranking upgrades or downgrades in Table 4-4. We find that only INTC is the one classified into “upgraded” stocks but delivers relatively negative excess returns, while the others all contribute to beautiful outperformances. As for those “downgraded” stocks, most of them have poor performances but only GE in 1999~1H00 and AAPL in 2H13 generate positive excess returns. The accumulated excess returns of the all “Downgraded” stock group are negative at -34% in total, which looks very bad comparing with that (+121.77%) of “Upgraded” group. The large deviation of total accumulated excess returns between Upgraded and 18. 25.

(26) Downgraded stocks indicates the ranking up/downgrade plays a key role in Portfolio’s performances. The underperformance of Portfolio A reflects the fact that the top market value stocks usually have no space to move forward but downgrade. The long-term reversal has been described as Momentum Crash in many studies, meaning the momentum strategy suffers from occasional drawdowns. Some literatures investigating the momentum crash argue that theses occurrences are concentrated on the market rebound. Because the momentum strategy is constructed based on the stocks’ past returns, it is likely to perform negatively by buying past winners and short past losers if the portfolios are formed after the market decline. I do find Portfolio A reports negative returns during Year 2000-2002, Year. 政治大. 2008-2009, when the market is reversed. But it also delivers extremely negative excess. 立. returns during Year 2012-2013 when the market (S&P 500) has been trending up; whist. ‧ 國. 學. Portfolio B generates strong positive returns at the same time. The argument cannot explain. ‧. why the No.1 market cap stock “cashes” during many of the observing periods but No.2 stock has always been a winner. I will discuss this issue from other point of views in the next. n. al. er. io. sit. y. Nat. section.. Ch. engchi. 19. 26. i Un. v.

(27) Table 4-2: Stocks’ 6-mth Returns of Portfolio A This table reports the different holding stock’s 6-mth returns of Portfolio A. The excess returns are the stocks’ 6-mth returns subtracting by that of S&P 500 index fund. Portfolio A Hol di ng Stock. 6-mth Return (%). Exces s Return. GE. MSFT. XOM. AAPL. S&P 500 Index Fund. GE GE GE GE MSFT MSFT MSFT GE GE GE GE MSFT MSFT GE GE GE GE GE GE XOM XOM XOM XOM XOM XOM XOM XOM XOM XOM XOM XOM AAPL AAPL XOM AAPL AAPL AAPL AAPL AAPL AAPL AAPL. 31.48 12.88 23.85 12.24 30.06 29.45 (31.48) (9.55) 1.69 (17.78) (27.52) (5.48) (0.81) 8.02 4.58 12.65 (5.07) 1.15 (5.96) 24.91 9.46 11.70 (5.93) (9.42) (12.43) (2.46) (16.31) 28.12 11.30 4.15 0.96 (8.87) (25.49) 12.01 15.95 18.78 13.63 (16.08) (9.18) 21.15 24.35. 11.89 2.97 7.11 3.41 18.96 22.24 (30.39) 0.15 8.24 (11.01) (14.10) 5.36 (11.46) (5.96) 1.66 6.80 (3.67) (3.32) (8.14) 13.60 3.24 14.51 6.54 20.07 (14.32) (23.66) (8.93) 6.29 6.35 9.05 (7.49) (13.50) (38.14) (3.12) 9.98 13.76 13.48 (15.12) (11.93) 14.44 16.17. 31.48 12.88 23.85 12.24 10.78 36.95 2.75 (9.55) 1.69 (17.78) (27.52) (16.18) 17.78 8.02 4.58 12.65 (5.07) 1.15 (5.96) 12.89 2.88 (3.16) (28.00) (39.30) (27.65) 29.10 (4.69) 26.84 3.12 (5.04) 16.36 0.72 10.48 20.87 (6.24) (3.84) 5.14 17.24 1.06 0.38 (14.53). 52.95 2.27 67.70 27.97 30.06 29.45 (31.48) (45.78) 68.30 (9.25) (17.43) (5.48) (0.81) 6.75 4.35 3.96 (7.04) 5.27 (10.90) 28.15 (1.31) 20.80 (22.72) (29.33) 22.27 28.23 (24.51) 21.30 (6.84) (0.15) 17.84 (12.68) 29.34 8.29 11.47 11.39 (4.95) 25.66 (7.77) 21.44 10.93. 25.00 (0.10) 16.65 2.45 5.47 4.46 (2.56) 10.75 0.50 (10.02) 4.12 (14.61) 2.78 14.17 8.32 15.42 12.11 (2.26) 9.22 24.91 9.46 11.70 (5.93) (9.42) (12.43) (2.46) (16.31) 28.12 11.30 4.15 0.96 1.15 4.39 12.01 (0.51) (8.17) (10.01) (6.31) 20.26 (3.71) (10.56). (31.74) (7.88) 118.56 42.70 13.12 122.00 1.88 (71.60) 56.30 (5.80) (19.09) (19.13) 33.00 12.12 52.27 97.91 14.32 95.30 (20.34) 48.14 43.85 62.31 (15.47) (49.03) 66.88 47.96 19.36 28.24 4.06 20.65 44.20 (8.87) (25.49) 41.48 15.95 18.78 13.63 (16.08) (9.18) 21.15 24.35. 19.59 9.91 16.74 8.83 11.10 7.21 (1.09) (9.70) (6.55) (6.77) (13.42) (10.84) 10.65 13.98 2.92 5.85 (1.40) 4.47 2.18 11.31 6.22 (2.81) (12.47) (29.49) 1.89 21.20 (7.38) 21.83 4.95 (4.90) 8.45 4.63 12.65 15.13 5.97 5.02 0.15 (0.96) 2.75 6.71 8.17. engchi. 20. 27. y. sit. er. n. Ch. ‧. io. al. 政治大. 學. ‧ 國. 立. Nat. 1H97 2H97 1H98 2H98 1H99 2H99 1H00 2H00 1H01 2H01 1H02 2H02 1H03 2H03 1H04 2H04 1H05 2H05 1H06 2H06 1H07 2H07 1H08 2H08 1H09 2H09 1H10 2H10 1H11 2H11 1H12 2H12 1H13 2H13 1H14 2H14 1H15 2H15 1H16 2H16 1H17. 6-mth Return (%). i Un. v.

(28) Table 4-3: Stocks’ 6-mth Returns of Portfolio B This table reports the different holding stock’s 6-mth returns of Portfolio B. The excess returns are the stocks’ 6-mth returns subtracting by that of S&P 500 index fund. Portfolio B. 6-mth Return (%). 1H97. KO. 6-mth Return (%) 29.22. 9.63. 29.22. 52.95. 31.48. 8.31. 25.00. (31.74). NA. 2H97. KO. (1.93). (11.84). (1.93). 2.27. 12.88. (0.93). (0.10). (7.88). NA. 9.91. 1H98. KO. 28.21. 11.47. 28.21. 67.70. 23.85. 5.52. 16.65. 118.56. NA. 16.74. Holding Stock. Excess Return. KO. MSFT. GE. INTC. XOM. AAPL. GOOGL. S&P 500 Index Fund 19.59. 2H98. MSFT. 27.97. 19.14. (21.64). 27.97. 12.24. 59.95. 2.45. 42.70. NA. 8.83. 1H99. GE. 10.78. (0.32). (7.46). 30.06. 10.78. 0.37. 5.47. 13.12. NA. 11.10. 2H99. GE. 36.95. 29.74. (6.05). 29.45. 36.95. 38.34. 4.46. 122.00. NA. 7.21. 1H00. GE. 2.75. 3.84. (1.39). (31.48). 2.75. 62.41. (2.56). 1.88. NA. (1.09). 2H00. INTC. (55.03). (45.33). 6.09. (45.78). (9.55). (55.03). 10.75. (71.60). NA. (9.70). 1H01. XOM. 0.47. 7.02. (26.15). 68.30. 1.69. (2.70). 0.50. 56.30. NA. (6.55). 2H01. MSFT. (9.25). (2.48). 4.78. 1H02 2H02. MSFT GE. (17.43) (16.18). (4.01) (5.34). 18.77 (21.71). 17.78. 7.13. 6.75. (7.23). 1H04. MSFT. 4.35. 2H04. MSFT. 1H05 2H05. 立. 5.86. 7.52. (10.02). (5.80). NA. (6.77). (17.43) (5.48). (27.52) (16.18). (41.91) (14.78). 4.12 (14.61). (19.09) (19.13). NA NA. (13.42) (10.84). 17.78. 33.65. 2.78. 33.00. NA. 10.65. 6.75. 8.02. 54.01. 14.17. 12.12. NA. 13.98. 1.43. (0.53). 4.35. 4.58. (13.88). 8.32. 52.27. NA. 2.92. (6.44). (12.29). (17.51). 3.96. 12.65. (15.25). 15.42. 97.91. NA. 5.85. XOM. 12.11. 13.51. 0.26. (7.04). (5.07). 11.24. 12.11. 14.32. 52.58. (1.40). XOM. (2.26). (6.73). (3.45). 5.27. 1.15. (4.07). (2.26). 95.30. 41.04. 4.47. 1H06. XOM. 9.22. 7.04. 6.72. (10.90). (5.96). (23.88). 9.22. (20.34). 1.08. 2.18. 2H06. GE. 12.89. 1.58. 12.16. 28.15. 12.89. 6.58. 24.91. 48.14. 9.81. 11.31. 8.41. (1.31). 2.88. 17.23. 9.46. 43.85. 13.51. 6.22. 17.32. 20.80. (3.16). 12.30. 11.70. 62.31. 32.29. (2.81). (39.30). sit. 2.88. (3.16). 1H08. GE. (28.00). 2H08. GE. 1H09 2H09. MSFT MSFT. 1H10. (0.35). (15.30). (22.72). (28.00). (19.43). (9.81). (12.91). (29.33). (39.30). (31.75). 22.27 28.23. 20.38 7.03. 6.01 18.77. 22.27 28.23. (27.65) 29.10. 12.89 23.26. MSFT. (24.51). (17.13). 2H10. AAPL. 28.24. 6.41. 1H11. AAPL. 4.06. (0.89). 2H11. AAPL. 20.65. 25.55. 3.98. io. (15.53). (5.93). (15.47). (23.87). (12.47). (9.42). (49.03). (41.56). (29.49). 66.88 47.96. 37.04 47.06. 1.89 21.20. er. GE GE. y. (3.34). Nat. 1H07 2H07. ‧. (0.81). 9.35. ‧ 國. GE MSFT. (17.78). 學. 1H03 2H03. 政治大 (9.25). (12.43) (2.46). n. a(12.07) l C (24.51) (4.69) (4.66)n i v(16.31) 31.23 h21.30 8.12 U 28.12 e n g26.84 c h i 5.37 2.31 (6.84) 3.12 11.30 (0.15). (5.04). 19.36. (28.23). (7.38). 28.24. 33.49. 21.83. 4.06. (14.75). 4.95. 9.43. 4.15. 20.65. 27.55. (4.90) 8.45. 1H12. AAPL. 44.20. 35.75. 11.75. 17.84. 16.36. 9.90. 0.96. 44.20. (10.19). 2H12. XOM. 1.15. (3.48). (7.28). (12.68). 0.72. (22.63). 1.15. (8.87). 21.95. 4.63. 1H13. XOM. 4.39. (8.26). 10.65. 29.34. 10.48. 17.51. 4.39. (25.49). 24.46. 12.65. 2H13. AAPL. 41.48. 26.35. 2.99. 8.29. 20.87. 7.12. 12.01. 41.48. 27.30. 15.13. 1H14. XOM. (0.51). (6.48). 2.54. 11.47. (6.24). 19.05. (0.51). 15.95. 4.24. 5.97. 2H14. XOM. (8.17). (13.19). (0.33). 11.39. (3.84). 17.44. (8.17). 18.78. (9.24). 5.02. 1H15. XOM. (10.01). (10.16). (7.08). (4.95). 5.14. 16.19. (10.01). 13.63. 1.77. 0.15. 2H15 1H16. GOOGL GOOGL. 44.07 (9.57). 45.03 (12.32). 9.51 5.52. 25.66 (7.77). 17.24 1.06. 13.27 (4.79). (6.31) 20.26. (16.08) (9.18). 44.07 (9.57). (0.96) 2.75. 2H16. GOOGL GOOGL. 12.64. 5.93. (8.54). 21.44. 0.38. 10.58. (3.71). 21.15. 12.64. 6.71. 17.32. 9.14. 8.18. 10.93. (14.53). (6.98). (10.56). 24.35. 17.32. 8.17. 1H17. 21. 28.

(29) Table 4-4: Holding stocks in Portfolio B by ranking upgrade/downgrade This table reports the 7 names that have been selected into Portfolio B in different periods by ranking upgrades or downgrades. We find that only INTC is the one classified into “upgraded” stocks but delivers relatively negative excess returns, while the others all contribute to beautiful outperformances. The accumulated excess returns of the all “Downgraded” stock group are negative at -34% in total, which looks very bad comparing with that (+121.77%) of “Upgraded” group. Upgraded from lower ranks Period 97-1H98. Downgraded from top rank. Accum. Excess Return. Stocks KO. 9.26. 2H98. MSFT. 19.14. 2H00. INTC. 1H05-2H06. Period. Accum. Excess Return. Stocks GE. 33.26. 2H01-1H02. MSFT. (6.49). (45.33). 2H03-2H04. MSFT. (18.09). XOM. 13.81. 2H06-2H08. GE. (27.46). 1H09-1H10. MSFT. 10.27. 2H10-1H12. AAPL. 2H15-1H17. GOOGL. 立 47.78. 2H12-1H13 治 XOM 政2H13 大AAPL 1H14-1H15. 26.35. XOM. 121.77. (29.83) (34.00). 155.77. ‧. ‧ 國. (11.74). 學. Subtotal. 66.83. 99-1H00. Deviation. sit. y. Nat. 4.3 Why Does Portfolio B Outperform?. n. al. er. io. We have discussed the market value ranking upgrade/downgrade have substantial. i Un. v. effect on the performances. In this section, I will go more details in analyzing why the. Ch. engchi. “upgraded “ 2nd largest market value stocks has huge opportunities to be outperformed. Explanation 1: fundamentals yet to peak Here I define the operating margin as the fundamental signal to compare with the changes of market value rankings. The stocks with upward rankings usually have improving op. margin in supporting the market values expansion. When the stocks are climbed to No.2 from No.3, their positive fundamentals tend to sustain a longer time after the upgrades. By examining the six stocks in Portfolio B that are upgraded from the lower ranks as I classified in Table 4-5, I found five of which appear to have such phenomenon. For example, MSFT is upgraded to No.2 from No.3 during 2H97. The company’s Op margin was improved from 30% 22. 29.

(30) in 1997 to over 40% by year 2000. Portfolio B enjoys excess returns in 2H98, which is one year before the fundamentals peak (See Appendix). XOM was also upgraded from No.3 to No.2 during 2003~2004 while its Op margin was seeing a 4-year enrichment toward 2007. The stock is selected into Portfolio B by end-2004 that generates excess returns of 13.8% during 1H05~2H06, also right before the margin peaks in 2007. AAPL was moving forward from No.3 to No.2 in 1H10, accompanied by the expanding Op. margins that peaked in 1H12. The stock in Portfolio B generated remarkably high total excess returns of 66.8% during 2H10~1H12. GOOGL, the stock upgraded to Portfolio B from. 政治大. the No.3 ranking by 2015, though hard to be explained its fundamental by Op. margin, also. 立. deliver beautiful excess returns of 47.8% during 2H15~1H17 that we believe the fundamental. ‧ 國. 學. is yet to peak.. ‧. In a word, I found the ranking tends to change ahead of the fundamental alteration. By picking the No.2 stocks, especially those “upgraded” ones, we might have higher. y. Nat. er. io. sit. probabilities to ensure the investments to earn excess returns given they are more likely growth stocks which yet to peak. Their volatilities could be lower than the small caps.. n. al. Ch. Explanation 2: Behavioral finance. engchi. i Un. v. It looks like the 2nd largest stocks tend to have more spaces than the top 1 company in share price expansions to deliver higher excess returns. It can also be explained by the investors’ behaviors. A moving-up stock could be considered as the fact that most investors believe it is still undervalued. The expectation in believing the share prices are not peaking draws inflows from the global investors to buy the stocks for Alpha returns. The more investors buying the story, the higher excess returns of the stocks. When the stock finally moves forward to the No.1 position, a number of investors may lose their interests in holding it as they believe the stock might have been peaked. The ranking is an anchor against which investors judge the intrinsic values of stocks. The 23. 30.

(31) investor’s anchoring bias come from the observations that many No.1 stocks go into to a downturn cycle after share price peaks. There will be more and more investors have such considerations if the No.1 1 stock cannot create excess returns as they expected. Underweight the largest cap stock would be one the strategies in portfolio construction, which leads to the price run-down of the No.1 stock. The actions from the bias-prone investors will trigger passive funds (ETFs)’s demand in selling the stocks due to price corrections. The No.1 stocks therefore become the loser in the portfolios.. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. 24. 31. i Un. v.

(32) 5. Conclusion This paper aims to investigate the outcome of single stock selection strategy by only picking the very large-market-value stocks. In my study, the strategy based on picking the second largest market value stocks proves to generate significant excess returns in the U.S. market. The ideas about finding outperforming large-cap stocks not only contradict the size effect theory that small-cap stocks tend to yield higher risk-adjusted returns, but also against the Modern Portfolio Theory (MPT) which highly emphasizes the importance of. 政治大. diversification to reduce risks. I therefore firstly examine the arguments of size effect theory. 立. by comparing the performance between large caps (S&P 100) and small caps (Russell 2000). ‧ 國. 學. over the period of 31/Jan/1990-2017/5/30. I find the size effect may exist in many of the cases, but not as significant as it was before. This supports my initial ideas of picking single. ‧. sock from the large marker value stock given it is too difficult to select it from the small caps,. Nat. sit. y. regardless the increasing uncertainties of performances. The literatures I collected also. n. al. er. io. suggest the more concentrated portfolios usually accompany with expanding excess returns.. i Un. v. The investment strategy I propose in the Chapter 3 is a stock picking simulation among. Ch. engchi. the No.1 and No.2 largest market value stocks, which has similar concept as Momentum Strategy that buying past winners and selling past losers . By comparing the two portfolios that are regularly rebalanced every 6 month, I find the 2nd largest market cap stock generated amazing excess returns while the No.1 stocks are poorly underperformed. In this paper, I provide explanations that No.1 stocks are more likely to experience Momentum Crash than No.2 stocks due to investor’s anchoring bias as they believe the No.1 stock might have been peaked. The ranking is an anchor against which investors judge the stock values. Another explanation comes from the observation of fundamental changes. Many of the No.2 stocks are in the process of growing stage, that the fundamentals are yet to peak 25. 32.

(33) comparing with the No.1. Since the rankings tend to change accompanied with share prices ahead of the fundamental alteration, it is likely to earn profits by buying No.2 stocks, especially those upgraded from the lower ranks. Although my research suggests buying the second largest market value stock for excess returns is workable, I believe this strategy shall be good to applied to other large caps except the No.1 stock as long as their rankings are upgrading. However, it would be difficult for individual investors to figure out the ranking changes in most of the time. Top three or four might be possible to identify but those ranked after the 5th stocks are unlikely to do so. Therefore, my advisories for implementation would be picking those super large cap stocks. 政治大. with faster-than-peers’ market value expansion but staying away the top one.. 立. The regular rebalances in my strategy are designed to be a mechanism in preventing the. ‧ 國. 學. misbehaviors of investors. The stocks should be dropped if they fail to maintain their. ‧. positions in the market value rankings, meaning they are too weak to be held in the portfolios, as what Momentum Strategy suggests. On the other hand, we should keep those. y. Nat. er. io. sit. with an upgrading position as they are apparently outperforming than peers. This is how momentum trading investors usually do. The difference would be my strategy has much. n. al. Ch. i Un. v. longer holding periods and less trading costs given the market value ranking of the large caps. engchi. does not change as often as the moving average of share prices. Still, my results do not prove the strategy can work in other markets, or how No.3 and No.4 stocks perform in the similar simulation. It also might be the case that other, yet unidentified factors would be able to explain the results. However, my conclusion is that the circumstantial evidences in the literatures and in my study points toward the fact that large market value stocks are capable of earing better excess returns. Investor should find their own strategies that are best appropriate for them. Further research could involve using a broader scope of markets and stocks to test whether the No.2 or No.3 stock outperform in many of the cases. For individual investors 26. 33.

(34) who are looking for single stock selections, the tactical use of this strategy can offer some performances.. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. 27. 34. i Un. v.

(35) 6. References Banz, Rolf W. (1981). “The Relationship Between Return and Market Value of Common Stocks,” Journal of Financial Economics, 9, 3-18. Byun, Suk Joon and Jeon , Byoung Hyun (2015). “Momentum crashes and an investor’s anchoring bias”. Working paper. Cohen, Randolph, Christopher Polk, and Bernhard Silli. (May 2010). “Best Ideas.” Working paper, London School of Economics. Daniel, Kent D., and Tobias J. Moskowitz. (2014). “Momentum crashes”, Working paper. 政治大. Dijk , Mathijs A. van. (2006) “Is size dead? A review of the size effect in equity returns”.. 立. Journal of Banking & Finance, 5-6.. ‧ 國. 學. Dowen, Richard J., and Bauman, W. Scott. (Spring 1986) “The relative importance of size, P/E, and neglect”. Journal of Portfolio Management.. ‧. Fama, Eugene F., and Kenneth R. French. (March 1995 ) “Size and book-to-market factors in. Nat. sit. y. earnings and returns”. Journal of Finance.. n. al. er. io. Fama, Eugene F., and Kenneth R. French. (June 1992) “The cross-section of expected stock returns”. Journal of Financial Economics.. Ch. engchi. i Un. v. Gompers, P.A., Metrick, A. (2001) “Institutional investors and equity prices”. Quarterly Journal of Economics 116, 229–259. Grey, Wesley R. (2014). “Does Size Effect Exist? Probably”. Blog.alphaarchitect. com. Jegadeesh, Narasimhan, and Sheridan Titman. (1993), “Returns to buying winners and selling losers: Implications for stock market efficiency”, Journal of Finance 48, 65-91. Lustig, Ivan L. and Philip A. Leinbach. (May–June 1983) “The Small Firm Effect.”Financial Analysts Journal 39, 46–49. Siegel, Jeremy (1994). “Stocks for the Long Run”, Chapter 6, page 95-96 Sonders , Liz Ann. (2017) “Big Machine: Why Large Caps Are Likely to Outperform”, Advisor 28. 35.

(36) Perspectives, page 1-3, March 14, 2017 Yeung, Danny, Paolo Pellizzari, Ron Bird, and Sazali Abidin. (2012) “Diversification versus Concentration …and the Winner is?” Working paper series, University of Technology Sydney.. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. 29. 36. i Un. v.

(37) Ch. engchi. 37. 30. 0 1 2 3 4 5 6 7 8 9 10. KO. Figure 7-3 Ranking changes of KO Jan-17. i Un. Jan-16. Figure 7-2 Op margin of GE. Jan-15. y. -10. Jan-14. sit. GE. Jan-12 Jan-13. Jan-11. Jan-10. Jan-09. Jan-07 Jan-08. Jan-06. Jan-05. er. al. Jan-04. 立. Jan-02 Jan-03. 10. Jan-01. ‧ 國 15. Jan-00. n. Jan-99. 1H96 1H97 1H98 1H99 1H00 1H01 1H02 1H03 1H04 1H05 1H06 1H07 1H08 1H09 1H10 1H11 1H12 1H13 1H14 1H15 1H16 1H17. -5. ‧. io. Jan-97 Jan-98. 5. 學. Nat. Jan-96. Jan-17. Jan-16. Jan-15. Jan-14. Jan-13. Jan-12. Jan-11. Jan-10. Jan-09. Jan-08. Jan-07. Jan-06. Jan-05. Jan-04. Jan-03. Jan-02. Jan-01. Jan-00. Jan-99. Jan-98. Jan-97. Jan-96. 7. Appendix. 0. 1. 2. 3. 4. 5. 6. 7 GE. Figure 7-1 Ranking change of GE. 政治大. 0. v.

(38) al Jan-01. Ch engchi. 38. 31 Jan-10. Jan-09. Jan-08. Jan-07. Jan-06. Jan-05. i Un. er. Jan-12. Jan-11. v. 2. 3. 4. 5. 6. MSFT. Figure 7-6 Ranking changes of MSFT Jan-17. Jan-16. Jan-15. Figure 7-5 Op Margin of XOM. sit y. XOM. Jan-14. 4. Jan-13. 立. Jan-04. 12. Jan-03. 14. Jan-02. 8. Jan-00. 10. Jan-99. n. 1. Jan-98. 1H96 1H97 1H98 1H99 1H00 1H01 1H02 1H03 1H04 1H05 1H06 1H07 1H08 1H09 1H10 1H11 1H12 1H13 1H14 1H15 1H16 1H17. 0. ‧. io. 0. Jan-97. ‧ 國 6. 學. Nat. Jan-96. 0. 1. 2. 3. 4. 5. 6. 7 XOM. Figure 7-4 Ranking changes of XOM. 政治大. 2. Jan-17. Jan-16. Jan-15. Jan-14. Jan-13. Jan-12. Jan-11. Jan-10. Jan-09. Jan-08. Jan-07. Jan-06. Jan-05. Jan-04. Jan-03. Jan-02. Jan-01. Jan-00. Jan-99. Jan-98. Jan-97. Jan-96.

(39) n. Jan-98 Jan-99. Jan-97. io. 0 1 2 3 4 5 6 7 8 9 10. al Ch. 39. 32. engchi Jan-09 Jan-10. Jan-08. Jan-06 Jan-07. Jan-04 Jan-05. Jan-03. Jan-01 Jan-02. Jan-00. i Un. er. v. AAPL. Figure 7-9 Ranking changes of AAPL Jan-17. Jan-15 Jan-16. Jan-14. Jan-12 Jan-13. INTC. y. Figure 7-8 Ranking changes of INTC. sit. Jan-11. ‧ 國立 ‧. 0 1 2 3 4 5 6 7 8 9 10. 學. Nat. Jan-96. Jan-17. Jan-16. Jan-14 Jan-15. Jan-13. Jan-12. Jan-11. Jan-10. Jan-08 Jan-09. Jan-07. Jan-06. Jan-05. Jan-04. Jan-02 Jan-03. Jan-01. Jan-00. Jan-99. Jan-98. Jan-96 Jan-97. 1H96 1H97 1H98 1H99 1H00 1H01 1H02 1H03 1H04 1H05 1H06 1H07 1H08 1H09 1H10 1H11 1H12 1H13 1H14 1H15 1H16 1H17. 45 40 35 30 25 20 15 10 5 0. MSFT. Figure 7-7 Op margin of MSFT. 政治大.

(40) 4. 5. 7. 8. 9. 學. 2. 立. GOOGL. io. n. al Ch engchi. 33. 40. ‧. Nat. y. 3. sit. 1. er. 6. ‧ 國. i Un. v. Figure 7-11 Ranking changes of GOOGL. 0. 政治大 Jan-17. Jan-16. Jan-15. Jan-14. Jan-13. Jan-12. Jan-11. Jan-10. Jan-09. Jan-08. Jan-07. Jan-06. Jan-05. Jan-04. Jan-03. Jan-02. Jan-01. Jan-00. Jan-99. Jan-98. Jan-97. Jan-96 1H96 1H97 1H98 1H99 1H00 1H01 1H02 1H03 1H04 1H05 1H06 1H07 1H08 1H09 1H10 1H11 1H12 1H13 1H14 1H15 1H16 1H17. 30 25 20 15 10 5 0 -5 -10 -15 -20. AAPL. Figure 7-10 Op Margin of AAPL.

(41) 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. 1. 41. i Un. v.

(42)