結合ESG揭露分數與價格動能策略之分析 - 政大學術集成

全文

(1)國立政治大學財務管理研究所碩士學位論文. 結合 ESG 揭露分數與價格動能策略之分析 An Analysis of Combining ESG-Scores and Price Momentum. 立. Strategies 治政大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. 指導教授：陳鴻毅. 研究生：蘇鈺婷. i n U. v. 博士. 撰. 中華民國一Ｏ八年六月. DOI:10.6814/NCCU201900078.

(2) 摘要本研究主要探討彭博資料庫中提供的 ESG 揭露分數結合價格動能策略是否能產生超額報酬。實證結果顯示買進過去輸家且公司治理分數(G)最低的組合(P1G1)並賣出過去贏家且 G 分數最高的組合(P5G5)，持有 12 個月可獲得 0.93％的顯著報酬。此結果表示過去為輸家且公司治理揭露得分最低的公司可能容易被低估股價，而得分最高的贏家可能會被高估。為避免金融海嘯造成動能策略失靈，我們近一步將樣本切分為 2010 年至 2017 年，並發現買進 P1ESG1 的小公司並賣出 P5ESG5 的小公司，持有 9 或 12 個月，分別可以獲得 13.73% 和 11.60%的顯著報酬。因此，我們推斷小型公司股票的效率. 政治大. 較低，即使它有 ESG 分數揭露，也很容易被低估或高估。當好消息發生時，ESG 得分. 立. 高的小公司可能被高估，相反地，當壞消息發生時，ESG 得分低的小公司很容易被低估。. Nat. n. al. er. io. sit. y. ‧. ‧ 國. 學. 關鍵字:價格動能;動能策略;企業社會責任;企業社會責任揭露. Ch. engchi. i n U. v. I. DOI:10.6814/NCCU201900078.

(3) Abstract In this study, we try to investigate whether the ESG disclosure scores provided in the Bloomberg database combined with price momentum strategies can generate excess returns. The empirical results show that buying the loser with the lowest G-Score (P1G1) and selling the winner with the highest G-Score in the past (P5G5), you can get a significant excess returns of 0.93% for 12 months. This result indicates that companies that used to be losers in the past and that have the lowest scores in corporate governance disclosures may be easily undervalued, while winners with the highest scores may be overvalued. In order to avoid the failure of the. 政治大. price momentum strategy caused by the financial crisis, we further divided the sample into 2010. 立. to 2017, and found that buying P1ESG1 of small size firm, while selling P5ESG5 of small size. ‧ 國. 學. firm, and hold for 9 or 12 months, significant excess returns of 13.73% and 11.60% were obtained. Therefore, we infer that the efficiency of small company stocks is low, even if it has. ‧. ESG-score disclosed, it is easy to be underestimated or overestimated. Small companies with. y. Nat. sit. high ESG-Score may be overvalued when good news happens. Conversely, small companies. n. al. er. io. with low ESG-Score can easily be underestimated when bad news occurs.. Ch. engchi. i n U. v. Keywords: price momentum; momentum strategy; ESG; ESG disclosure. II. DOI:10.6814/NCCU201900078.

(4) CONTENTS List of Tables ...................................................................................................................... IV List of Figures ...................................................................................................................... V 1 Introduction .......................................................................................................................1 2 Literature Review ..............................................................................................................4 2.1 Price momentum strategy .......................................................................................4 2.2 Relationship between momentum strategy, financial statement and return.........5 2.3 Relationship between ESG, financial statement and return ..................................7 3 Data and Sample Descriptions ..........................................................................................9. 政治大. 3.1 Data .........................................................................................................................9 3.2 Sample descriptions ................................................................................................9 3.3 Price momentum strategy ..................................................................................... 10 3.4 ESG strategy ......................................................................................................... 11 3.5 ESG combined price momentum strategy ........................................................... 12. 立. ‧ 國. 學. 4 Empirical Results ............................................................................................................. 13. ‧. er. io. sit. y. Nat. 4.1 Price momentum profit ......................................................................................... 13 4.2 ESG strategy ......................................................................................................... 13 4.3 ESG-momentum strategy ..................................................................................... 15 4.4 E-momentum strategy, S-momentum strategy and G-momentum strategy ....... 16 4.5 ESG-momentum strategy on small size firms during 2010 to 2017 .................... 17. al. n. v i n 5 Future Research and ConclusionC.................................................................................... 19 hengchi U. References ........................................................................................................................... 21. III. DOI:10.6814/NCCU201900078.

(5) List of Tables 1 Summary statistics……………………………………………..…………….……..……….23 2 Returns to price momentum strategies……………………………………….……….……..25 3 Returns to ESG strategies………………………………………………………...…………26 4 Returns to ESG-momentum strategies……………………………………….……………...30 5 Returns to ESG-momentum portfolios……………………………………………………...34 6 Contributions of E, S and G in ESG-momentum strategies………………….………..……..36 7 Returns to small size ESG-momentum portfolios…………………………….………….….39. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. i n U. v. IV. DOI:10.6814/NCCU201900078.

(6) List of Figures 1 Distribution of ESG-Scores…………………………………………………………………41. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. i n U. v. V. DOI:10.6814/NCCU201900078.

(7) 1 Introduction The ultimate goal pursued by all investors in the financial market is to obtain excess returns (or abnormal returns, the actual rate of return minus the expected rate of return). For many centuries, scholars, investors and financial researchers are exploring how to find a way to get excess returns from the market in a stable and long-term way through a systematic or regular portfolio, strategy or mechanism. In addition to studying how to effectively obtain excess returns, scholars explore and empirically analyze through different data frequencies, market patterns, financial indicators and. 政治大. other factors. The purpose is to examine investment strategies in different contexts through. 立. different methods and angles. The compensation results are expected to establish a long-term. ‧ 國. 學. and effective profit-making portfolio model under the same risks.. Basically, the efficiency market hypothesis in Malkiel and Fama (1970) believe that the. ‧. stock price will reflect all relevant information, with the information quickly responding to the. y. Nat. sit. market, investors have no way to use this information to continuously receive excess returns.. n. al. er. io. While investors believe market inefficient think that they could receive excessive compensation,. i n U. v. and it is due to the asymmetry information or the incomplete interpretation of information in the short term.. Ch. engchi. The price momentum strategy in Jegadeesh and Titiman (1993) is commonly known as the method of chasing high returns portfolio (winner) and selling the low one (loser), and there are many followers who believe in market inefficient and behavior finance follow it. Such investors believe that stock prices generally have a lack of reaction, and will continue for a period of time, causing the winners maintain high returns, and losers maintain low or negative returns in several months. Therefore, it is possible to obtain a strategy by buying a winner combination with a short-selling of losers. In the past, there have been many papers that extended the price momentum strategy and 1. DOI:10.6814/NCCU201900078.

(8) added factors from financial statements to try and find ways to get excess returns, such as trying price-earnings momentum in Chordia and Shivakumar (2006) or considering price, revenue and earnings at the same time in Chen, Chen, Hsin, and Lee (2014). However, this study hopes to combine the price momentum strategy with the ESG (Environmental, Social, Corporate Governance) score to make a different extension from previous researches and papers. In recent years, ESG issues have been heatedly discussed in countries around the world and even in Taiwan. The question of company managers is that if the better ESG-Score is, the higher the value of the company will be, and the relative question. 治政大 investing in companies with high ESG-Score. 立. of investors is if them could expect a long-term stable compensation by supporting or even. In addition, when investors study whether ESG will increase the value of the company, the. ‧ 國. 學. information collected or the completeness and transparency of the score will also have an. ‧. impact on the outcome. Such as Tamimi and Sebastianelli (2017) explore the state of S&P 500. sit. y. Nat. companies’ disclosure transparency by analyzing their Bloomberg ESG-Scores and Fatemi,. io. decreases it.. er. Glaum, and Kaiser (2018) find that ESG disclosures increase firm value but its weakness also. al. n. v i n This study will combine the price C hmomentum strategyUwith ESG-Scores in the Bloomberg engchi. database. The data of the U.S. listed companies from 2004 to 2017 will be divided into winners (with the highest past returns) and losers (with the lowest past returns) by different formation period and holding period, including 3-month, 6-month, 9-month and 12-month. The dataset will also be divided into high or low ESG-Scores (or E-Score, S-Score, and G-Scores respectively) independently, to explore what type of stock portfolio to buy can be profitable, and if the ESG-momentum strategy that buying winner with high ESG-Score and selling loser with low ESG-Score could earn excess return. We also extend our discussion to size factor, exploring whether stocks of small companies are more inefficient in Fama and French (1993), so its stock price would easily be 2. DOI:10.6814/NCCU201900078.

(9) underestimated or overestimated with good news or bad news occurring, and then we could use momentum strategy to beat the market. At last, we hope this study provides investors with future investment strategies, and also the reference factors for stock selection, which can be made by using the historical returns and ESG-Scores.. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. i n U. v. 3. DOI:10.6814/NCCU201900078.

(10) 2 Literature Review 2.1 Price momentum strategy The price momentum strategy refers to an investment strategy that buys a winner’s portfolio with a relatively high rate of return over the past period, and at the same time shorts loser’s portfolio, and holds for a period of time. Based on this result, it is inferred that the stock price of the market is underreacted, so it is possible to obtain excess returns through the price momentum strategy. According to the efficiency market hypothesis, this market is a weak efficiency market, so it can use past information to obtain excess compensation.. 政治大. The following are studies of the sample by different scholars through different stock. 立. markets and periods, and their empirical analyses based on different research methods. The. ‧ 國. 學. conclusions of various scholars are combined to infer that the stock markets of some countries in the past has a significant excess returns using price momentum strategy.. ‧. Jegadeesh and Titiman (1993) use the US stock market data from 1965 to 1989 to show. y. Nat. sit. that stock returns exhibit momentum behavior at intermediate horizons. It finds that a self‐. n. al. er. io. financing strategy that buys the top 10% (winner) and sells the bottom 10% (loser) of stocks. i n U. v. ranked by returns during the past 6 months, and holds the positions for 6 months, produces. Ch. engchi. considerable abnormal profits of 1% per month.. Jegadeesh and Titman (2001) use additional 9-year data from 1990 to 1998 sample period, and find that Jegadeesh and Titman (1993) momentum strategies is still a profitable way and that past winners outperform past losers by nearly the same magnitude as in previous 1965 to 1989 period. Rouwenhorst (1998) studies the stock markets of 12 countries from 1980 to 1995 and found that these stock markets had a mid-term price momentum effect with significant excess returns, and it calls that return continuation. It also reveals that return continuation is negatively related to firm size, while it is not limited to small firms. 4. DOI:10.6814/NCCU201900078.

(11) Moskowitz and Grinblatt (1999) use 20 value‐weighted industry portfolios, formed for every month from July 1963 to July 1995, and show a strong momentum effect in industry components of stock returns. It concludes that industry momentum investment strategies, which buy stocks from past winning industries and sell stocks from past losing industries, appear highly profitable. . George and Hwang (2004) compare the momentum strategies to Jegadeesh and Titman (1993), present a challenge to the view that markets are semi-strong form efficient, and they use the similar grouping way, the winner (loser) portfolio in momentum strategy is the equal-. 治政大 for the 52‐week high strategy return to which the stock belongs, and the winner (loser) portfolio 立. weighted portfolio of the top (bottom) 30% of stocks ranked by the value‐weighted industry. is the equal-weighted portfolio of the 30% of stocks with the highest (lowest) ratio of current. ‧ 國. 學. price to 52‐week high. It concludes that a readily available piece of information, the 52‐week. ‧. high price, explains a large portion of the profits from momentum strategy, and consider that. sit. y. Nat. the 52‐week high is a better predictor of future returns than are past returns.. n. al. er. io. 2.2 Relationship between momentum strategy, financial statement and return. i n U. v. Many scholars try to obtain excess return from financial statements, and the most famous. Ch. engchi. one is from Fama and French (1992) using companies’ book-to-market ratio as a variable to capture the financial distress, and thus the subsequent returns represent a fair compensation for risk. In addition to only invest in firms with good financial statement information, some scholars also combine information from financial statement and momentum strategy of buying high / good firms and selling low / bad at the same time to construct a more profitable investment way with zero-cost. Dichev and Piotroski (2001) extend the study base on Fama and French (1992) and show that the mean annual return earned by a high book-to-market ratio investor can be gained by at 5. DOI:10.6814/NCCU201900078.

(12) least 7.5% and within the portfolio of high book-to-market ratio firms, the benefits to financial statement analysis are concentrated on small and medium-sized firms and low share turnover companies. In addition, an investment strategy that buys expected winners and shorts expected losers generates a 23% annual return during 1976 to 1996, and the strategy appears to be robust with time series with controlling for alternative investment strategies. Chan, Jegadeesh, and Lakonishok (1996) use the US stock market data to construct an earnings-momentum investment strategy and found that the investment portfolio formed by stocks with good earnings in the past 6 months earned 8.8% during the next 6-month period.. 政治大. Chordia and Shivakumar (2006) conduct time-series and cross-sectional asset pricing tests,. 立. they ﬁnd that the systematic component of earnings momentum captures price momentum.. ‧ 國. 學. Besides, the predictive power of past returns is subsumed by a zero-investment portfolio, which is longing on stocks with high earnings surprises with shorting on stocks with low earnings. ‧. surprises at the same time.. y. Nat. io. sit. Chen, Chen, Hsin, and Lee (2014) consider price, earnings, and revenue as momentum. n. al. er. strategies. The study found a significant return on the revenue-momentum strategy, holding a. Ch. i n U. v. three-month, six-month, and nine-month winner portfolio, could earn average monthly returns. engchi. of 0.94%, 0.93%, and 0.84%, respectively. Besides, they also found that these earnings momentum strategy have a significant positive return for a holding period of 3 to 12 months. In Chen, Lee, and Shih (2016), extending the previous paper and conduct a ratio of liquidity with buying volume to liquidity and selling volume (BOS ratio). They show that the BOS momentum strategy can enhance the profits of momentum strategy. Then, they combine two fundamental indicators, the FSCORE and the GSCORE into momentum strategy. The empirical results show that the combined investment strategy includes stocks with a huge information content that the combined one cannot include in time, because the market reaction is insufficient, and the abnormal investment behavior is continued to be more and more. Therefore, the 6. DOI:10.6814/NCCU201900078.

(13) combined investment strategy generates significantly higher returns and outperforms momentum strategy. 2.3 Relationship between ESG, financial statement and return As the topic of ESG is hot these decades, more and more scholars, investors and firm managers want to know if firms would do well by doing good, so there are many theses focus on the returns of ESG funds or social responsibility investment. Bauer, Koedijk, and Otten (2005) conclude that there is no significant difference in compensation between ESG funds and general funds, and ethical funds have experienced a. 政治大. catch-up period to traditional ones. In addition, this study analyzes the characteristics of ESG. 立. funds in various regions, such as UK and German ethical funds are heavily exposed to small. ‧ 國. 學. caps, while ethical funds in the US, on the other hand, invest more in large caps compared to their conventional peers.. ‧. Friede, Busch and Bassen (2015) combine the findings of about 2,200 individual studies,. y. Nat. sit. and said roughly 90% of studies show a non-negative ESG-CFP (Corporate financial. n. al. er. io. performance) relation, and the large majority of studies reports positive findings. It concludes. i n U. v. that there would be a positive ESG impact on corporate financial position, and it is stable over time.. Ch. engchi. As for how ESG disclosure quality impacts on investment returns, Gelb and Strawser (2001) propose an alternative explanation that firms disclose themselves willingly because it is the socially responsible thing to do. It indicates a positive relationship between corporate social responsibility and disclosure level. That is, firms that engage in socially responsive activities provide more informative and extensive disclosures than companies that are less focused on advancing social responsibilities. In addition, Fatemi, Glaum, and Kaiser (2018) find that ESG strengths increase firm value and ESG weaknesses decrease it, and disclosure plays a crucial moderating role by reducing the negative effect of weaknesses and also the positive effect of 7. DOI:10.6814/NCCU201900078.

(14) strengths. Thesis focuses on the ESG-Score on Bloomberg is like Tamimi and Sebastianelli (2017), exploring the state of S&P 500 companies’ disclosure transparency by analyzing their Bloomberg ESG score. They conclude that there are significant differences in disclosure transparency on both Social and Governance dimensions, found between certain industry sectors. The results also present that large-cap companies have significantly higher ESG disclosure scores than mid-cap companies.. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. i n U. v. 8. DOI:10.6814/NCCU201900078.

(15) 3 Data and Sample Descriptions 3.1 Data We collect data from CRSP (The Center for Research in Security Prices) to obtain monthly holding period return, stock price, share outstanding, share codes and exchange codes of U.S. listed firms, and only common stock and firms list on Nasdaq, American Stock Exchange and New York Stock Exchange are included. The sample period is from January 2004 to December 2017. As for ESG-Scores, we collect yearly data from Bloomberg. If the company is public,. 政治大. Bloomberg monitors its environmental, social, and governance (ESG) performance, and even. 立. if the company does not issue its sustainability report. Bloomberg researchers crunch thousands. ‧ 國. 學. of companies’ ESG disclosure performance into one number as its ESG disclosure score with company published data and news items, like annual reports, sustainability reports, press release. ‧. and third-party research. 0 to 100 is the range of the Bloomberg ESG disclosure score, and it. y. Nat. sit. doesn’t measure performance, but transparency. The more information disclosed, the higher the. n. al. er. io. disclosure score. Take year 2014 as example, 1 was the lowest score, and 89 was the highest. i n U. v. score. The rating item details cover the range from energy, emission, waste, women on the board,. Ch. engchi. independent directors, workforce accidents, sector specific, and so on. At last, sample firms we use should have corresponding ESG-Score and monthly returns in each formation month. It is considered that the ESG-Score will be delayed as the financial statement. Therefore, when we correspond to the monthly rate of return, we assume that we will not receive the ESG-Score information until next July. Besides, the observations of ESG-Scores in year 2004 are only 6, so we exclude them in the ESG strategies and ESG-momentum strategies for the difficulty of making groups. 3.2 Sample descriptions. 9. DOI:10.6814/NCCU201900078.

(16) Table1 presents the summary statistics for major variables, and Panel A including the total period distributions of ESG-Score, E-Score, S-Score, G-Score and Monthly returns. There are 3525 firms in our sample, the mean score of ESG-Score, E-Score, S-Score and G-Score are respectively 17.01, 20.82, 15.88 and 51.00, and we could find that the mean of G-Score is much higher than other scores. In addition, the minimum and maximum value of scores are about 1 and 77 to 87, and E-Score has a largest standard deviation, 17.45, while G-Score has the lowest one, 5.82, during the whole period. Panel B shows detailed score distributions in each year, and we have 23772 early ESG-. 治政大 is concentrated on 15.52 to more than 2000 observations every year. The mean of ESG-Score 立. Score in the dataset. We could see that ESG-Score observations increase quickly since 2010,. 22.79, E-Score is from 16.31 to 29.84, S-Score is from 12.36 to 19.44 and G-Score is from. ‧ 國. 學. 50.34 to 54.46. Obviously, G-Score has the highest average score than others, and S-Score has. n. al. <Insert Table1 here >. Ch. engchi. er. io. sit. y. Nat. concentrated.. ‧. the lowest one. At the same time, G-Score has the lowest standard deviation, its scores are. i n U. v. We could also observe the trend of ESG-Score, E-Score, S-Score and G-Score from 2005 to 2017 on Figure.1, and only S-Score could be concluded to an apparently upward trend.. <Insert Figure1 here >. 3.3 Price momentum strategy Base on Jegadeesh and Titman (1993), we construct portfolios “winner” and “loser”. First, defining the beginning of each month, t, from January 2004 to December 2017 and rank stocks 10. DOI:10.6814/NCCU201900078.

(17) based on their past J-month returns, this is so called formation period. The second step is to make groups, and in this step we adjust grouping numbers from Jegadeesh and Titman (1993) 10 groups to form 5 equal-weighted portfolios, with the top 20% portfolio is called “winner” means the highest ranking-period returns and the bottom 20% portfolio is called “loser” means the lowest ranking-period returns. Third, in each month t, using the price momentum strategy to long the winner portfolio and short the loser portfolio at the same time to come up a zero-cost investment strategy, and then holds this position for K months, which means the holding periods of this strategies, and. 治政 stocks are ranked based on their past 12-month returns, and大 then held for 12 months, and in this 立 additionally closes out the position formed in month t + K. For example, with J=12, K=12,. thesis, we conduct the strategies with J={3, 6, 9, 12}, K={3, 6, 9, 12} to analyze.. ‧ 國. 學. The portfolios are with overlapping holding periods, so in every month, t, there are. ‧. respectively K separate winner and loser portfolios, and the return of winners and losers are. sit. y. Nat. calculated as the equal-weighted average of month t returns from K separate winner (losers). io. er. portfolios. To calculate the profits of momentum strategy, there are two ways we can use, one is buy-and-hold return, and the other is rebalanced return, which is rebalanced every month to. al. n. v i n maintain equal weights. We choose C the rebalance return in this thesis. hengchi U 3.4 ESG strategy. We use the same grouping way as price momentum strategy, form 5 equally weighted portfolios, among them, using the top 20% ESG-Score group (ESG5) and the bottom 20% group (ESG1), like “winners” and “losers”, to conduct the ESG strategies with J={3, 6, 9, 12} and K={3, 6, 9, 12}, and calculate the portfolios return of following months. However, the observations of ESG-Scores in year 2004 are only 6, so we exclude them in the ESG strategies for the difficulty of making groups. The data period is from 2005 to 2017. Following Fama and French (1996) and Jegadeesh and Titman (2001), we also use the 11. DOI:10.6814/NCCU201900078.

(18) capital asset pricing model (CAPM) and Fama French 3 factor model (FF3) to examine whether the ESG strategy returns could be explained by pricing factors, including the risk-free rate, SMB and HML from Kenneth French’s website. 3.5 ESG combined price momentum strategy In this two-way grouping strategy, we independently rank the past returns and ESG-Scores from the formation period J={3, 6, 9, 12} to obtained 25-group portfolios with K={3, 6, 9, 12}, which are like P1(Loser)ESG1(Low), P1ESG2, P1ESG3, …, P5(Winner)ESG5(High), and then tracking their average returns in the following holding periods. However, the observations of. 政治大. ESG-Scores in year 2004 are only 6, so we exclude them in the ESG-momentum strategies for. 立. the difficulty of making groups. The data period is from 2005 to 2017.. ‧ 國. 學. Table4 shows the returns and CAPM and FF3 adjusted returns of P1ESG1, P5ESG5 and the zero-cost strategy with longing P5ESG5 and shorting P1ESG1. Table5 shows the detailed. ‧. returns to ESG-momentum portfolios of each group.. n. er. io. sit. y. Nat. al. Ch. engchi. i n U. v. 12. DOI:10.6814/NCCU201900078.

(19) 4 Empirical Results 4.1 Price momentum profit Base on Jegadeesh and Titman (1993), we construct portfolios “winner” and “loser” every month, and count the returns oh them in the following month. In Table2, P2, P3, and P4 usually have a higher significant level up to 0.05, while P1 and P5 seldom have significant returns with J={3, 6, 9, 12}, K={3, 6, 9, 12}. Second, we don’t observe price momentum profit by buying the winner portfolios (P5) and selling the loser (P1), even try the different formation period and holding period with J={3, 6, 9, 12}, K={3, 6, 9, 12}.. 政治大 simulated the price momentum立 strategy, unlike Jegadeesh and Titman (1993) data period is. We speculate the possible reason is that the data period we conduct is too short to be. ‧ 國. 學. about 25 years. In addition, maybe the momentum crash occurred in Daniel and Moskowitz (2016), during the financial crisis from 2007 to 2009. It means that although price momentum. ‧. strategy has positive average returns across asset classes, in panic economic states with high. sit. y. Nat. market volatility it would easily to become loss. During the economic recession, the prices of. er. io. losers embody a high premium, when the market rebounds, the losers would experience strong. al. n. v i n C situation can also explain before. On the other hand, maybe this h e n g c h i U why portfolio P2 to P4 have. gains, and result in a momentum crash as the price momentum strategy shorted these losers. significant returns while winners and losers are not.. <Insert Table2 here >. 4.2 ESG strategy Table3 presents returns to ESG strategies, and here we use the same grouping way as price momentum strategy, form 5 equally weighted portfolios, among them, using the top 20% ESGScore group (ESG5) and the bottom 20% group (ESG1), like “winners” and “losers”, to conduct 13. DOI:10.6814/NCCU201900078.

(20) the ESG strategies. We also apply E-Score, S-Score and G-Score separately as different investment strategies with J=12 and K={3, 6, 9, 12}. In addition to return, we use the capital asset pricing model (CAPM) and Fama French 3 factor model (FF3) to examine whether the ESG strategy returns could be explained by pricing factors, including the risk-free rate, SMB and HML from Kenneth French’s website.. The two regressions are below: 𝛾𝛾𝑝𝑝,𝑡𝑡 − 𝛾𝛾𝑓𝑓,𝑡𝑡 = 𝛼𝛼𝑝𝑝 + 𝛽𝛽𝑝𝑝 �𝛾𝛾𝑚𝑚,𝑡𝑡 − 𝛾𝛾𝑓𝑓,𝑡𝑡 � + 𝑒𝑒𝑡𝑡. (1). 政治大. 𝛾𝛾𝑝𝑝,𝑡𝑡 − 𝛾𝛾𝑓𝑓,𝑡𝑡 = 𝛼𝛼𝑝𝑝 + 𝛽𝛽𝑝𝑝1 �𝛾𝛾𝑚𝑚,𝑡𝑡 − 𝛾𝛾𝑓𝑓,𝑡𝑡 � + 𝛽𝛽𝑝𝑝2 𝑆𝑆𝑆𝑆𝑆𝑆𝑡𝑡 + 𝛽𝛽𝑝𝑝3 𝐻𝐻𝐻𝐻𝐻𝐻𝑡𝑡 + 𝑒𝑒𝑡𝑡. 立. (2). ‧ 國. 學. Where 𝑟𝑟𝑝𝑝,𝑡𝑡 is the return on portfolio P, 𝑟𝑟𝑓𝑓,𝑡𝑡 is the interest rate on 1-month treasury bill, 𝛾𝛾𝑚𝑚,𝑡𝑡. is the return on market index, 𝑆𝑆𝑆𝑆𝑆𝑆𝑡𝑡 is the size factor of Fama French 3 factor model (small. ‧. minus large firms’ stocks), and 𝐻𝐻𝐻𝐻𝐻𝐻𝑡𝑡 is the book-to-market factor of Fama French 3 factor. model (high minus low book-to market stocks).. sit. y. Nat. n. al. er. io. It’s easily observed that using ESG-Score, E-Score, S-Score and G-Score could explain. i n U. v. portfolio returns most of time, especially the holding period is 6-month and 9-month, and even. Ch. engchi. consider pricing factors from capital asset pricing model and Fama French 3 factor model. Among them, S2(the bottom 20% to 40% Social score) and S4 with 9-month holding period have 0.01 significant level of Fama French 3 factor alpha of 1.49% and 1.33%, and G1(the lowest 20% Governance score) with 6-month and 9-month holding period have 0.01 significant level of Fama French 3 factor alpha of 1.38% and 1.41%. However, it’s no significant abnormal returns from ESG-Strategy by buying stocks with high ESG-Score and selling stocks with low ESG-Score, neither the E-Score, S-Score and GScore.. 14. DOI:10.6814/NCCU201900078.

(21) <Insert Table3 here >. 4.3 ESG-momentum strategy Table4 presents returns to ESG-momentum strategies. We consider both the ESG-Score of each stock and the past return with J={3, 6, 9, 12}, K={3, 6, 9, 12}. Panel A and Panel B conclude the results of ESG-momentum strategy with 3-month formation period, and it could be observed that holding P1ESG1 (loser with lowest ESG score) for 6 or 9 months could have respectively 1.53% and 1.42% return, after the adjustment of Fama French 3 factor model, and. 政治大. the zero-cost strategy that buying P5ESG5 (winner with highest ESG score) and selling. 立. P1ESG1 with 12-month holding period would have a negative return 0.98%.. ‧ 國. 學. Panel C and Panel D conclude the results of ESG-momentum strategy with 6-month formation period, and it could be observed that holding P1ESG1 (loser with lowest ESG score). ‧. for 9 months could have 1.72% return, after the adjustment of Fama French 3 factor model, it. y. Nat. sit. comes to 2.05%. The zero-cost strategy that buying P5ESG5 (winner with highest ESG score). n. al. er. io. and selling P1ESG1 with 9-month holding period would have a negative return 1.48%.. i n U. v. As for Panel E to Panel H with J={9, 12}, K={3, 6, 9, 12}, it shows that the ESG-. Ch. engchi. momentum strategy by buying P5ESG5 and selling P1ESG1 would not gain a significant return. We speculate the causes of this result are two, which are momentum crash and size factor. First, momentum crash in Daniel and Moskowitz (2016) usually happened in financial crisis, the price of losers embodies a high premium during the recession, when the market rebounds, the losers would experience strong gains, and results in a momentum crash as the price momentum strategy shorted these losers before. Second, big companies reduce the effect of momentum strategy in Fama and French (2012) indicate that spreads in average momentum returns decrease from smaller to bigger stocks, and the stock we selected from Bloomberg is based on their ESG disclosure score, so our data may affect by the phenomenon that firms 15. DOI:10.6814/NCCU201900078.

(22) disclose their movements on ESG on their own initiative are usually large in Tamimi and Sebastianelli (2017).. <Insert Table4 here >. Table5 reports the detailed excess returns of 25 groups we divided into, which are the pairs of 5 past-return groups and 5 ESG-Score groups with 12-month formation period. We find that P2, P3 and P4 with different ESG-Score usually have significant returns, for. 治政大 However, there is no significant excess returns from ESG-momentum strategy by buying 立. example, P2ESG1 have 0.01 significant level of 1.45% return with 3-month holding period.. stocks with high ESG-Score winner and selling stocks with low ESG-Score loser. Only buying. ‧ 國. 學. P5ESG1 and selling P5ESG5 could have 1.04% excess return, it may indicate that it’s easy to. n. al. <Insert Table5 here >. Ch. engchi. er. io. sit. y. Nat. disclosure.. ‧. underestimate winner with low ESG disclosure, and overestimate winner with high ESG. i n U. v. 4.4 E-momentum strategy, S-momentum strategy and G-momentum strategy Table6 shows contributions of E-Score, S-Score and G-Score in ESG-momentum strategies. We form the groups from the past return and these 3 score respectively with J={12}, K={3, 6, 9, 12}. Presenting way as Table4, in Panel A and Panel B, we report the return and adjusted return by using CAPM and Fama French 3 factor model of P1E1 (loser with the lowest E score) and P5E5 (winner with the highest E score) and also the zero-cost strategy by buying P5E5 and selling P1E1. Panel C and Panel D are the analyses of S-Score and the past return, and Panel E and Panel F are for G-Score and the past return. 16. DOI:10.6814/NCCU201900078.

(23) We find that P5E5 has a significant excess return of 1.41% with J={12}, K={9}, but Emomentum strategy is not effective. Second, S-momentum strategy is still not effective, but P1S1 has a significant excess return respectively 1.90% and 1.65% with J={12}, K={6, 9}. At last, P1G1 has the higher level of significance return with J={12}, K={6, 9}, and they are respectively 1.82% and 1.88%. In addition, the strategy of G-momentum strategy works with J={12}, K={12} as we buy P1G1 and sell P5G5 and hold for 12 months could earn 0.93% at average, and it shows that the loser with the lowest governance score may easily be underestimated, while the winner with the highest governance score may be overestimated.. 立. 治政 <Insert Table6 here > 大. ‧ 國. 學. 4.5 ESG-momentum strategy on small size firms during 2010 to 2017. ‧. As we want to remove the impact from financial crisis period that reducing the power of. y. Nat. sit. price momentum and investigate if momentum strategy could earn by small size (price. n. al. er. io. multiples shares outstanding) firm, we adjust the data used period to 2010 to 2017 and make. i n U. v. group of firms with ESG-Score into 5 groups, and only use the smallest 10% size to analyze.. Ch. engchi. <Insert Table7 here >. In Table7, we show the detailed returns with 12-month formation period from 25 group we divided into, and they are all small firms. Panel B presents that small size ESG-momentum with 6-month holding period and we could observe that strategy with buying P4ESG5 and selling P4ESG1 would have a significantly negative return of 7.17%. It could mean that the market underestimated the stock of ESG1, so it has a rebound after 6 months, or more accurately, its lack of disclosure reduces 17. DOI:10.6814/NCCU201900078.

(24) the efficiency of market, so it takes higher premium than ESG5. Panel C presents with 9-month holding period, and there are 2 strategies work. First, longing P5ESG5 and shorting P1ESG1 would have a significantly negative return 13.73%, and secondly, even longing P5ESG5 and shorting P5ESG1, which are both winner portfolios, would have a significantly negative return 14.36%. Base on the results above, we conclude that market inefficient occurs while the firm size is small, so the return of stocks is easy to be underestimated or overestimated. When the good news happened, and the stock with high ESGScore, these small companies could be overestimated, and conversely, when the bad news. 治政大 and shorting P5ESG1 would In Panel D with 12-month holding period, longing P5ESG5 立. happened, these small companies with low ESG-Score could easily be underestimated.. have a significantly negative return 11.60%, and longing P5ESG5 and shorting P5ESG1, which. ‧ 國. 學. are both winner portfolios, would have a significantly negative return 11.91%. The result is like. ‧. Panel C, revealing the small size market inefficient, so our strategy works.. n. er. io. sit. y. Nat. al. Ch. engchi. i n U. v. 18. DOI:10.6814/NCCU201900078.

(25) 5 Future Research and Conclusion In the past centuries, there have been many papers extended the price momentum strategy and added factors from financial statements to try and find ways to get excess returns, such as Chordia and Shivakumar (2006) trying price-earnings momentum or Chen, Chen, Hsin, and Lee (2014) considering price, revenue and earnings at the same time. However, this study combines the price momentum strategy with the ESG (Environmental, Social, Corporate Governance) score on Bloomberg to make a different extension from previous researches and papers. In this thesis, we focus on if considering ESG-Score and the past return. 政治大 For future researches, they 立could extend this thesis to use ESG performance scores as. at the same time could form a better momentum strategy.. ‧ 國. 學. variables, not only ESG disclosure scores, maybe it would receive another opinion. Besides, grouping and the selection of the data period would also be a big issue, such as observing. ‧. whether the January effect will affect the profits of the momentum strategies, or trying to find. er. io. Scores at the same time would earn excess returns.. sit. y. Nat. whether constructing a strategy of buying and selling matching firms with and without ESG-. al. n. v i n C h 2004 to December Bloomberg in the U.S. market from January 2017, we could not earn a engchi U For conclusion, as we use the full data sample of all companies with ESG-Score on. significantly excess return from price momentum strategy, ESG strategy and ESG-momentum strategy. The result is like Bauer, Koedijk, and Otten (2005) conclude that there is no significant difference in compensation between ESG funds and general funds. We infer that the reason why momentum invalid is because only company disclosures its ESG movement on annual reports, sustainability reports, press release and third-party research would be scored at Bloomberg. It causes the database we use for momentum strategy failed to work, because the data we selected are usually big, high transparent, and maybe they are doing good thing, so they are willing to disclosure more, is like Gelb and Strawser (2001) indicate a. 19. DOI:10.6814/NCCU201900078.

(26) positive relationship between corporate social responsibility and disclosure level. Even Emomentum and S-momentum strategy do not have a significant excess return, and only buying P1G1 (loser with the lowest G score) and selling P5G5 (winner with the highest G score) and hold for 12 months could earn 0.93%. On the other hand, momentum strategy usually works in the inefficient market with information asymmetry, and Tamimi and Sebastianelli (2017) present that large-cap companies have significantly higher ESG disclosure scores than mid-cap companies, so we try to use the small size (price multiples shares outstanding) data, which is the smallest 10% size in our. 政治大. dataset, to conduct the ESG-momentum strategy, and hope it could let the momentum strategy work.. 立. We could find that buying small size P1ESG1 (loser with the lowest ESG score) and selling. ‧ 國. 學. small size P5ESG5 (winner with the highest ESG score) and hold for 9 months and 12 months. ‧. could respectively earn 13.73% and 11.60%. Therefore, we infer that the small size company. sit. y. Nat. stocks are less inefficient in Fama and French (1993), and even it has ESG-Score, it’s easy to. io. er. be underestimated or overestimated. When the good news happened, and the small companies with high ESG-Score could be overestimated, and conversely, when the bad news happened,. al. n. v i n and the ESG-Score is low, these small easily be underestimated. C companies h e n g ccould hi U. 20. DOI:10.6814/NCCU201900078.

(27) References Bauer, Rob, Kees Koedijk, and Roger Otten, 2005, International evidence on ethical mutual fund performance and investment style, Journal of Banking and Finance 29, 1751-1767. Chan, Louis KC, Narasimhan Jegadeesh, and Josef Lakonishok, 1996, Momentum Strategies, Journal of Finance 51, 1681-1713. Chen, Hong-Yi, Lee, Cheng-Few, and Shih, Wei K., 2016, Technical, fundamental, and combined information for separating winners from losers, Pacific-Basin Finance Journal 39, 224-242.. 治政大 of Banking and Finance 38, momentum drive or ride earnings or price momentum, Journal 立. Chen, Hong-Yi, Chen, Sheng-Syan, Hsin, Chin-Wen, and Lee, Cheng-Few, 2014, Does revenue. 166-185.. ‧ 國. 學. Chordia, Tarun, and Lakshmanan Shivakumar, 2006, Earnings and price momentum, Journal. ‧. of Financial Economics 80, 627-656.. sit. y. Nat. Daniel, Kent, and Tobias J. Moskowitz, 2016, Momentum crashes, Journal of Financial. io. er. Economics 122, 221-247.. Dichev, Ilia D., and Joseph D. Piotroski, 2001, The long‐run stock returns following bond. n. al. Ch. ratings changes, Journal of Finance 56, 173-203.. engchi. i n U. v. Fama, Eugene F., and Kenneth R. French, 1992, The cross‐section of expected stock returns, Journal of Finance 47, 427-465. Fama, Eugene F., and Kenneth R. French, 1993, Common risk factors in the returns on stocks and bonds, Journal of Financial Economics 33, 3-56. Fama, Eugene F., and Kenneth R. French, 1996, Multifactor explanations of asset pricing anomalies, Journal of Financial Economics 51, 55-84. Fama, Eugene F., and Kenneth R. French, 2012, Size, value, and momentum in international stock returns, Journal of Financial Economics 105, 457-472. Fatemi, Ali, Martin Glaum, and Stefanie Kaiser, 2018, ESG performance and firm value: The 21. DOI:10.6814/NCCU201900078.

(28) moderating role of disclosure, Global Finance Journal 38, 45-64. Friede, Gunnar, Timo Busch, and Alexander Bassen, 2015, ESG and financial performance: aggregated evidence from more than 2000 empirical studies, Journal of Sustainable Finance & Investment 5, 210-233. Gelb, David S., and Joyce A. Strawser, 2001, Corporate social responsibility and financial disclosures: An alternative explanation for increased disclosure, Journal of Business Ethics 33, 1-13. George, Thomas J., and Chuan‐Yang Hwang, 2004, The 52-week high and momentum investing,. 治政大 to buying winners and selling Jegadeesh, Narasimhan, and Sheridan Titman, 1993, Returns 立 Journal of Finance 59, 2145-2176.. losers: Implications for stock market efficiency, Journal of Finance 48, 65-91.. ‧ 國. 學. Jegadeesh, Narasimhan, and Sheridan Titman, 2001, Profitability of momentum strategies: An. ‧. evaluation of alternative explanations, Journal of Finance 56, 699-718.. sit. y. Nat. Malkiel, Burton G., and Fama, Eugene F., 1970, Efficient capital markets: A review of theory. io. er. and empirical work, Journal of Finance 25, 383-417.. Moskowitz, Tobias J., and Mark Grinblatt, 1999, Do industries explain momentum, Journal of. n. al. Finance 54, 1249-1290.. Ch. engchi. i n U. v. Rouwenhorst, K. Geert, 1998, International momentum strategies, Journal of Finance 53, 267284. Tamimi, Nabil, and Rose Sebastianelli, 2017, Transparency among S&P 500 companies: an analysis of ESG disclosure scores, Management Decision 55, 1660-1680.. 22. DOI:10.6814/NCCU201900078.

(29) Table 1 Summary statistics for major variables This table reports summary descriptive statistics for all variables, including ESG-Score, E-Score, S-Score, G-Score, monthly return and firm numbers. The sample period is from 2004 to 2017. Panel A is the summary statistics during the whole period. Panel B classifies variables to different year.. Panel A. Summary Statistics Obs.. Mean. Std.. Median. ESG-Score. 23772. 17.0112. 10.1203. 13.2200. E-Score. 4907. 20.8287. 17.4478. 15.5000. 治政 77.2700 大 0.7800 82.1700. S-Score. 14074. 15.8820. 13.1855. 8.7700. 86.6700. G-score. 23652. 51.0091. 5.8219. 51.7900. 1.0000. 85.7100. Monthly Return. 271469. 0.0128. 0.1150. 0.0105. -0.8784. 3.7954. 0.8800. ‧ 國. 立1.0000. ‧. io. sit. y. Nat. n. al. er. 3525. Max. 學. # of firms. Min. Ch. engchi. i n U. v. 23. DOI:10.6814/NCCU201900078.

(30) Panel B. Distribution of ESG scores ESG-Score. E-Score. S-Score. G-Score. Year. Obs.. Mean. Median. Std.. Mean. Median. Std.. Mean. Median. Std.. Mean. Median. Std.. 2005. 141. 19.0328. 14.4600. 9.3504. 19.0044. 15.5000. 12.5432. 13.7223. 8.7700. 12.4780. 51.4963. 51.7900. 7.0551. 2006. 237. 21.9951. 17.3600. 10.8852. 20.3244. 17.8300. 13.2463. 15.8952. 8.7700. 14.5093. 53.0908. 51.7900. 6.5964. 2007. 878. 17.2703. 14.0500. 9.1471. 16.3131. 12.4550. 14.2605. 12.8141. 8.7700. 12.8093. 51.2682. 51.7900. 5.5222. 2008. 1249. 17.2245. 14.0500. 9.4661. 17.2588. 2009. 1316. 17.5305. 14.0500. 10.3294. 2010. 2463. 15.5246. 11.9800. 8.8966. 立 19.5920. 2011. 2656. 15.5477. 11.9800. 9.1429. 2012. 2758. 16.1896. 12.7200. 2013. 2791. 16.6507. 2014. 2806. 2015. 12.4117. 50.9839. 51.7900. 6.3438. 8.7700. 12.9108. 51.1487. 51.7900. 6.4155. 13.9500. 16.9398. 13.5591. 8.7700. 12.9895. 50.3770. 48.2100. 5.3134. 21.0587. 17.0500. 17.4812. 14.3673. 8.7700. 12.9563. 50.4267. 48.2100. 5.2566. 9.7505. 22.0864. 17.0500. 17.9493. 15.4645. 8.7700. 13.1224. 50.3442. 48.2100. 5.6252. 12.8100. 10.1592. 22.6046. 17.0500. 17.9910. 16.6931. 14.0400. 13.0132. 50.5534. 48.2100. 5.8723. 17.0541. 12.9200. 10.3478. 22.1887. 16.2800. 18.1735. 17.2791. 14.0400. 13.1233. 50.9220. 48.2100. 5.6933. 2817. 17.3663. 13.2200. 10.6021. 22.1811. 16.2800. 18.2214. 17.8268. 14.0400. 13.3236. 51.1211. 51.7900. 5.8620. 2016. 2640. 17.8411. 14.0400. 10.5877. 21.4862. 15.5000. 17.6073. 18.3195. 14.0400. 12.9059. 51.7287. 51.7900. 5.8013. 2017. 1014. 21.7770. 16.5300. 11.8757. 20.9249. 13.2850. 18.4085. 19.4401. 17.3650. 12.8396. 54.4647. 51.7900. 6.1472. 2005-2017. 23766. 17.0113. 13.2200. 10.1203. 20.8287. al. 15.5000. 17.4460. 15.8831. 8.7700. 13.1856. 51.0089. 51.7900. 5.8218. io. y. sit. Nat. n. Ch. er. 18.9746. ‧. ‧ 國. 8.7700. 學. 12.4500 治 15.1816 12.3666 政 13.9500 17.4583 大 13.0755. n engchi U. iv. 24. DOI:10.6814/NCCU201900078.

(31) Table 2 Returns to price momentum strategies This table reports the holding period returns of each groups classify by the past returns and price momentum strategy (P5-P1) during 2004 to 2017. P1 is the portfolio with the lowest ranking-period returns, P2 is the second lowest ranking-period returns, and so on. Panel A with 3-month and 6-month formation period, and Panel B with 9-month and 12-month formation period. ***, **, and * indicate statistical significance at the 1%, 5%, and 10 % level, respectively.. Panel A. Formation period of portfolio is 3 months and 6 months Formation period. 3 months. 6 months 政治大 12 months 3 months 6 months 9 months 立 0.0099 0.0089 0.0093 0.0096. Holding period. 3 months 6 months 9 months. P1(Loser). 0.0114*. P2. 0.0110** 0.0102* 0.0103**. 0.0100*. 0.0096* 0.0104** 0.0096*. P3. 0.0086*. 0.0095*. 0.0095** 0.0093* 0.0099** 0.0095**. P4. 0.0084* 0.0092** 0.0092*. P5(Winner). 0.0088*. 0.0086. 0.0083. 0.009. 0.0085*. 0.0077. 0.0087*. 0.0082. P5-P1. -0.0026. -0.0009. -0.0009. -0.0006. -0.0014. -0.0012. -0.0006. -0.002. 0.0096** 0.0094** 0.0097** 0.0098** 0.0102**. y 12 months. n. al. er. io. 9 months. ‧. Nat. Panel B. Formation period of portfolio is 9 months and 12 months Formation period. 0.0099*. sit. 0.0092*. 0.0102. 學. 0.0097*. 0.0092. ‧ 國. 0.0095. 12 months. i n U. v. P1(Loser). 0.0092. 0.0099. 0.0101. 0.0110*. Ch. P2. 0.0093*. 0.0089*. 0.0090*. 0.0094*. 0.0098*. 0.0098*. 0.0091*. P3. 0.0093* 0.0105** 0.0100** 0.0102** 0.0099** 0.0100** 0.0099**. 0.0091*. P4. 0.0098** 0.0101** 0.0113** 0.0115** 0.0104** 0.0117** 0.0123** 0.0119**. Holding period. 3 months 6 months 9 months 12 months 3 months 6 months 9 months 12 months. e n0.0106 g c h i 0.0112. 0.0101. 0.0096*. 0.0113. P5(Winner). 0.0076. 0.0078. 0.0074. 0.0082. 0.0075. 0.0069. 0.0081. 0.0072. P5-P1. -0.0016. -0.0021. -0.0027. -0.0028. -0.0026. -0.0037. -0.0031. -0.0041. 25. DOI:10.6814/NCCU201900078.

(32) Table 3 Returns to ESG strategies This table reports excess returns of each groups classify by ESG-Score, E-Scores, S-Scores, G-Scores and ESG strategy (ESG5-ESG1) during 2005 to 2017. ESG1 is the portfolio with the lowest ESG-Score, ESG2 is the second lowest ESG-Score, and so on. Panel A and Panel B based on ESG-Score, Panel C and Panel D based on E-Score, Panel E and Panel F based on S-Score and Panel G and Panel H based on G-Score. ***, **, and * indicate statistical significance at the 1%, 5%, and 10 % level, respectively.. 政治大 6-month holding period. Panel A. Investment strategy based on ESG-Score with 3-month and 6-month holding period. 立. 3-month holding period. FF3 Alpha. Return. CAPM Alpha. FF3 Alpha. ESG1(Low). 0.0105*. 0.0099*. 0.0099*. 0.0123**. 0.0150**. 0.0143**. ESG2. 0.0099*. 0.0102*. 0.0104*. 0.0117**. 0.0137***. 0.0134**. ESG3. 0.0098*. 0.0097*. 0.0100*. 0.0098*. 0.0134**. 0.0131**. ESG4. 0.0094*. 0.0096*. 0.0099*. 0.0089*. 0.0116**. ESG5(High). 0.0080*. 0.0078*. 0.0081*. 0.0079*. 0.0097**. ESG5-ESG1. -0.0025. -0.0021. -0.0018. -0.0044. -0.0053. 0.0110**. y. 0.0094**. io. sit. -0.0049. er. Nat. al. ‧. ‧ 國. CAPM Alpha. 學. Return. n. Panel B. Investment strategy based on ESG-Score with 9-month and 12-month holding period 9-month holding period. Ch. i n U. v. 12-month holding period. e Return n g c h i CAPM Alpha. Return. CAPM Alpha. FF3 Alpha. FF3 Alpha. ESG1(Low). 0.0126**. 0.0145**. 0.0142**. 0.0114*. 0.0093*. 0.0101*. ESG2. 0.0122**. 0.0141***. 0.0141***. 0.0117**. 0.0102*. 0.0107*. ESG3. 0.0101*. 0.0132**. 0.0131**. 0.0101*. 0.0091. 0.0097. ESG4. 0.0093*. 0.0117**. 0.0118**. 0.0074. 0.0062. 0.0072. ESG5(High). 0.0087*. 0.0103**. 0.0100**. 0.0081*. 0.0076*. 0.0086*. ESG5-ESG1. -0.0039. -0.0042. -0.0042. -0.0033. -0.0017. -0.0015. 26. DOI:10.6814/NCCU201900078.

(33) Panel C. Investment strategy based on E-Score with 3-month and 6-month holding period 3-month holding period. 6-month holding period. Return. CAPM Alpha. FF3 Alpha. Return. CAPM Alpha. FF3 Alpha. E1(Low). 0.0075. 0.0068. 0.0068. 0.0076. 0.0105*. 0.0099*. E2. 0.0097*. 0.0098*. 0.0100*. 0.0106**. 0.0129**. 0.0128**. E3. 0.0077. 0.0078. 0.0077. 0.0075. 0.0103*. 0.0100*. E4. 0.0078*. 0.0075. 0.0079. E5(High). 0.0069. 0.0060. E5-E1. -0.0006. -0.0008. 立 -0.0003. -0.0004. 0.0085* 0.0088*. -0.0017. -0.0011. 學. ‧ 國. 0.0065. 0.0092* 政0.0073治大 0.0072* 0.0088*. Panel D. Investment strategy based on E-Score with 9-month and 12-month holding period. 12-month holding period. ‧. 9-month holding period FF3 Alpha. Return. CAPM Alpha. FF3 Alpha. 0.0078. 0.0100*. 0.0103*. 0.0070. 0.0054. 0.0063. E2. 0.0113**. 0.0134**. 0.0130**. 0.0103*. 0.0094*. E3. 0.0070. 0.0089*. 0.0089*. 0.0072. 0.0062. E4. 0.0101**. 0.0119**. 0.0082*. 0.0078*. E5(High). 0.0067. 0.0082*. al. E5-E1. -0.0011. -0.0018. sit. io. 0.0116**. er. E1(Low). y. CAPM Alpha. Nat. Return. n. iv 0.0078* 0.0072 n C h 0.0081 e0.0011 -0.0025 n g c h i U0.0018. 0.0099* 0.0072 0.0089* 0.0078 0.0015. 27. DOI:10.6814/NCCU201900078.

(34) Panel E. Investment strategy based on S-Score with 3-month and 6-month holding period 3-month holding period. 6-month holding period. Return. CAPM Alpha. FF3 Alpha. Return. CAPM Alpha. FF3 Alpha. S1(Low). 0.0092*. 0.0085. 0.0087. 0.0095*. 0.0127. 0.0122**. S2. 0.0100*. 0.0106*. 0.0107*. 0.0111*. 0.0147***. 0.0140**. S3. 0.0094*. 0.0092*. 0.0093*. 0.0099*. 0.0127**. 0.0121**. S4. 0.0098*. 0.0100*. 0.0101*. S5(High). 0.0076*. 0.0072*. 0.0076*. S5-S1. -0.0016. -0.0013. -0.0025. 0.0117** 0.0080*. -0.0044. -0.0042. 學. ‧ 國. 立 -0.0011. 政0.0095*治 0.0123** 大 0.0070* 0.0083*. Panel F. Investment strategy based on S-Score with 9-month and 12-month holding period. 12-month holding period. ‧. 9-month holding period FF3 Alpha. Return. CAPM Alpha. FF3 Alpha. S1(Low). 0.0098*. 0.0116. 0.0115*. 0.0105*. 0.0084. 0.0090. S2. 0.0117*. 0.0150***. 0.0149***. 0.0103*. 0.0094*. S3. 0.0093. 0.0114*. 0.0114*. 0.0092. 0.0078. S4. 0.0107**. 0.0133***. 0.0089*. 0.0080. S5(High). 0.0074*. 0.0086*. al. S5-S1. -0.0024. -0.0030. sit. er. io. 0.0133***. y. CAPM Alpha. Nat. Return. n. iv 0.0084* 0.0066 n C h 0.0072* e-0.0033 -0.0031 n g c h i U-0.0018. 0.0106* 0.0087 0.0087* 0.0075* -0.0015. 28. DOI:10.6814/NCCU201900078.

(35) Panel G. Investment strategy based on G-Score with 3-month and 6-month holding period 3-month holding period. 6-month holding period. Return. CAPM Alpha. FF3 Alpha. Return. CAPM Alpha. FF3 Alpha. G1(Low). 0.0095*. 0.0091*. 0.0091*. 0.0122**. 0.0142***. 0.0138***. G2. 0.0155**. 0.0139*. 0.0143*. 0.0141*. 0.0196**. 0.0189**. G3. 0.0067. 0.0089. 0.0092. 0.0081. 0.0078. 0.0072. G4. 0.0090*. 0.0082. 0.0085. G5(High). 0.0086*. 0.0082*. 0.0084*. G5-G1. -0.0009. -0.0009. -0.0045. 0.0112** 0.0094*. -0.0046. -0.0044. 學. ‧ 國. 立 -0.0007. 政0.0086*治 0.0119** 大 0.0077* 0.0096*. Panel H. Investment strategy based on G-Score with 9-month and 12-month holding period. 12-month holding period. ‧. 9-month holding period FF3 Alpha. Return. CAPM Alpha. FF3 Alpha. 0.0126**. 0.0143***. 0.0141***. 0.0116**. 0.0098*. 0.0104*. G2. 0.0132. 0.0226**. 0.0227**. 0.0092. 0.0056. G3. 0.0082. 0.0078. 0.0074. 0.0088. 0.0082. G4. 0.0090*. 0.0112**. 0.0076. 0.0064. G5(High). 0.0092*. 0.0111**. al. G5-G1. -0.0034. -0.0032. sit. io. 0.0111**. er. G1(Low). y. CAPM Alpha. Nat. Return. n. iv 0.0111** 0.0078* n C h 0.0086* e -0.003 -0.003 n g c h i U-0.002. 0.0065 0.0088 0.0075 0.0087* -0.0017. 29. DOI:10.6814/NCCU201900078.

(36) Table 4 Returns to ESG-momentum strategies This table reports excess returns of the loser with the lowest ESG-Score (P1ESG1), the winner with the highest ESG-Score (P5ESG5) and the zero-cost strategy of buying P5ESG5 and selling P1ESG1 at the same time during 2005 to 2017. Panel A and Panel B based on 3-month formation period, Panel C and Panel D based on 6-month, Panel E and Panel F based on 9-month and Panel G and Panel H based on 12-month. ***, **, and * indicate statistical significance at the 1%, 5%, and 10 % level, respectively.. 政治大 6-month holding period. Panel A. 3-month formation period with 3-month and 6-month holding period. 立. 3-month holding period. Return. CAPM Alpha. FF3 Alpha. 0.0090. 0.0111. 0.0111. 0.0100. 0.0152**. 0.0153**. (1.63). (1.90). (1.90). (1.68). (2.60). (2.61). 0.0086. 0.0079. 0.0085. 0.0101*. 0.0129*. 0.0121*. (1.85). (1.62). (1.72). (1.97). (2.43). -0.0004. -0.0032. -0.0026. 0.0001. -0.0023. (-0.10). (-0.66). (-0.60). (0.13). (-0.72). n. 9-month holding period P1ESG1 P5ESG5 P5ESG5-P1ESG1. Ch. (2.27). y. -0.0032. sit. (-0.88). er. io. al. Panel B. 3-month formation period with 9-month and 12-month holding period. ‧. Nat. P5ESG5-P1ESG1. FF3 Alpha. ‧ 國. P5ESG5. CAPM Alpha. 學. P1ESG1. Return. i n U. v. 12-month holding period. eReturn n g c h i CAPM Alpha. Return. CAPM Alpha. FF3 Alpha. FF3 Alpha. 0.0118*. 0.0143*. 0.0142*. 0.0124. 0.0111. 0.0122. (2.05). (2.49). (2.46). (1.92). (1.68). (1.86). 0.0093. 0.0104*. 0.0095. 0.0027. 0.0018. 0.0025. (1.90). (2.07). (1.87). (0.59). (0.38). (0.51). -0.0025. -0.0039. -0.0047. -0.0097*. -0.0093. -0.0098*. (-0.62). (-0.97). (-1.18). (-2.01). (-1.92). (-2.19). 30. DOI:10.6814/NCCU201900078.

(37) Panel C. 6-month formation period with 3-month and 6-month holding period 3-month holding period FF3 Alpha. Return. CAPM Alpha. FF3 Alpha. 0.0079. 0.0095. 0.0101. 0.0101. 0.0143*. 0.0137*. (1.30). (1.50). (1.61). (1.66). (2.38). (2.26). 0.0081. 0.0073. 0.0083. 0.0068. 0.0082. 0.0074. (1.75). (1.47). (1.67). 0.0002. -0.0022. -0.0018. (0.19). (-0.70). 立 (-0.54). (1.56) 政(1.37) 治大 -0.0033 -0.0062 (-0.85). 學 ‧. CAPM Alpha. FF3 Alpha. Return. CAPM Alpha. FF3 Alpha. 0.0172*. 0.0213**. 0.0205**. 0.0126. 0.0100. 0.0118. (2.22). (2.73). (2.61). (1.79). (1.41). 0.0049. 0.0067. 0.0057. 0.0045. 0.0034. (1.04). (1.41). (1.00). (0.74). -0.0124. -0.0146*. al. (-1.82). (-2.18). er. (1.21). iv -0.0148* -0.0066 n C h -0.0081 e(-1.41) (-2.13) n g c h i U(-1.20). n. P5ESG5-P1ESG1. Return. io. P5ESG5. (-1.52). 12-month holding period. Nat. P1ESG1. -0.0063. (-1.49). Panel D. 6-month formation period with 9-month and 12-month holding period 9-month holding period. (1.40). y. P5ESG5-P1ESG1. CAPM Alpha. sit. P5ESG5. Return. ‧ 國. P1ESG1. 6-month holding period. (1.66) 0.0040 (0.86) -0.0078 (-1.44). 31. DOI:10.6814/NCCU201900078.

(38) Panel E. 9-month formation period with 3-month and 6-month holding period 3-month holding period FF3 Alpha. Return. CAPM Alpha. FF3 Alpha. 0.0086. 0.0111. 0.0108. 0.0162*. 0.0235**. 0.0222**. (1.37). (1.64). (1.61). (1.99). (2.95). (2.79). 0.0056. 0.0042. 0.0047. 0.0051. 0.0068. 0.0059. (1.13). (0.80). (0.88). -0.0031. -0.0069. -0.0061. (-0.88). (-1.52). 立 (-1.44). (1.41) 政(1.08) 治大 -0.0111 -0.0168* (-1.58). 學. FF3 Alpha. Return. CAPM Alpha. 0.0136. 0.0172*. 0.0162*. 0.0141. 0.0111. (1.73). (2.17). (2.03). (1.81). (1.40). 0.0071. 0.0088. 0.0081. 0.0074. 0.0062. (1.37). (1.68). (1.66). (1.35). -0.0065. -0.0084. al. (-0.97). (-1.27). FF3 Alpha 0.0124. er. (1.52). ‧. CAPM Alpha. iv -0.0082 -0.0049 n C h -0.0067 U e(-1.02) (-1.17) n g c h i (-0.86). n. P5ESG5-P1ESG1. Return. io. P5ESG5. (-2.32). 12-month holding period. Nat. P1ESG1. -0.0163*. (-2.46). Panel F. 9-month formation period with 9-month and 12-month holding period 9-month holding period. (1.20). y. P5ESG5-P1ESG1. CAPM Alpha. sit. P5ESG5. Return. ‧ 國. P1ESG1. 6-month holding period. (1.55) 0.0069 (1.51) -0.0055 (-1.01). 32. DOI:10.6814/NCCU201900078.

(39) Panel G. 12-month formation period with 3-month and 6-month holding period 3-month holding period FF3 Alpha. Return. CAPM Alpha. FF3 Alpha. 0.0132. 0.0136. 0.0131. 0.0156. 0.0237**. 0.0224**. (1.67). (1.63). (1.58). (1.79). (2.80). (2.66). 0.0033. 0.0029. 0.0032. 0.0097*. 0.0116**. 0.0107**. (0.72). (0.59). (0.64). -0.0099. -0.0107. -0.0099. (-1.53). (-1.65). 立 (-1.52). (2.74) 政(2.33) 治大 -0.0059 -0.0121 (-0.64). 學 ‧. FF3 Alpha. Return. CAPM Alpha. FF3 Alpha. 0.0156. 0.0200*. 0.0194*. 0.0154*. 0.0121. 0.0132. (1.90). (2.44). (2.35). (2.01). (1.56). 0.0071. 0.0085. 0.0078. 0.0073. 0.0066. (1.58). (1.88). (1.70). (1.51). -0.0086. -0.0115. al. (-1.06). (-1.53). (1.71). y. CAPM Alpha. iv -0.0116 -0.0056 n C h -0.0081 e(-1.10) (-1.48) n g c h i U(-0.85). n. P5ESG5-P1ESG1. Return. io. P5ESG5. (-1.31). 12-month holding period. Nat. P1ESG1. -0.0117. (-1.44). Panel H. 12-month formation period with 9-month and 12-month holding period 9-month holding period. (2.52). sit. P5ESG5-P1ESG1. CAPM Alpha. er. P5ESG5. Return. ‧ 國. P1ESG1. 6-month holding period. (1.72) 0.0073 (1.66) -0.0059 (-1.01). 33. DOI:10.6814/NCCU201900078.

(40) Table 5 Returns to ESG-momentum portfolios This table reports the detailed excess returns of 25 groups we divided into, which are the pairs of 5 past-return groups and 5 ESG-Score groups with 12-month formation period. The rightmost column and the bottommost column are the result of group subtraction. Panel A with 3-month holding period, Panel B with 6month holding period, Panel C with 9-month holding period and Panel D with 12-month holding period. ***, **, and * indicate statistical significance at the 1%, 5%, and 10 % level, respectively.. 政治大 ESG4 ESG5(High). Panel A. 12-month formation period with 3-month holding period. 立. ESG3. 0.0132. 0.0065. 0.0130. 0.0170*. 0.0088. -0.0073. P2. 0.0145***. 0.0142**. 0.0106*. 0.0103*. 0.0094*. -0.0048. P3. 0.0083. 0.0109**. 0.0090. 0.0102**. 0.0091**. 0.0011. P4. 0.0090*. 0.0089*. 0.0076. 0.0075. 0.0099**. 0.0009. P5(Winner). 0.0137*. 0.0053. 0.0089. 0.0036. 0.0033. 0.0005. -0.0012. -0.0041. -0.0134*. -0.0055. ESG4. ESG5(High). 0.0137*. 0.0106. v. -0.0052. 0.0076. -0.0032. ‧. -0.0104*. y. Nat. -0.0031. sit. P1(Loser). ‧ 國. ESG2. 學. ESG1(Low). al. ESG1(Low). ESG2. P1(Loser). 0.0156. 0.0157*. 0.0133. P2. 0.0105*. 0.0125**. 0.0125**. P3. 0.0110*. 0.0117**. 0.0091*. 0.0091*. 0.0088**. -0.0039. P4. 0.0062. 0.0120**. 0.0092*. 0.0091*. 0.0060. -0.0000. P5(Winner). 0.0108. 0.0097. 0.0041. 0.0064. 0.0097*. -0.0010. -0.0048. -0.006. -0.0092. -0.0073. -0.0009. 0.0042. n. ESG3. er. io. Panel B. 12-month formation period with 6-month holding period. Ch. e 0.0070 ngchi. i n U. 34. DOI:10.6814/NCCU201900078.

(41) Panel C. 12-month formation period with 9-month holding period ESG1(Low). ESG2. ESG3. ESG4. ESG5(High). P1(Loser). 0.0156*. 0.0150*. 0.0132. 0.0166*. 0.0057. -0.0088. P2. 0.0112*. 0.0150***. 0.0112*. 0.0086. 0.0094*. -0.0017. P3. 0.0111**. 0.0086. 0.0129**. 0.0075. 0.0085**. -0.0028. P4. 0.0118*. 0.0099*. 0.0071. 0.0096**. 0.0088*. -0.0029. P5(Winner). 0.0123*. 0.0121*. 0.0030. -0.0033. -0.0029. -0.0102. 立. 0.0071 政0.0085治大 -0.0081 0.0014. P3. 0.0080. 0.0073. P4. 0.0115**. 0.0117**. 0.0074. 0.0019. -0.008. -0.0163. P5(Winner). 0.0120. 0.0155**. 0.0059. 0.0081. 0.0111**. 0.0044. 0.0071*. 0.0095*. 0.0072. 0.0087**. 0.0061. 0.0112*. 0.0073. al. 0.0007. -0.0047. -0.0023. Ch. n engchi U. iv. -0.0052 0.0020 0.0007. y. 0.0164**. 0.0105. sit. 0.0062. 0.0084. er. P2. ‧ 國. 0.0182**. ESG5(High). ‧. 0.0154*. ESG4. n. P1(Loser). ESG3. io. ESG2. Nat. ESG1(Low). 0.0037. 學. Panel D. 12-month formation period with 12-month holding period. -0.0051. -0.0029 -0.0001 0.0051. 35. DOI:10.6814/NCCU201900078.

(42) Table 6 Contributions of E, S and G in ESG-momentum strategies This table reports excess returns of the loser with the lowest E-Score (P1E1), the winner with the highest E-Score (P5E5) and the zero-cost strategy of buying P5E5 and selling P1E1 at the same time, so do the S-Score and G-Score with 12-month formation period. Panel A and Panel B based on E-momentum. strategies, Panel C and Panel D S-momentum strategies, Panel E and Panel F based on G-momentum strategies. ***, **, and * indicate statistical significance at the 1%, 5%, and 10 % level, respectively.. 政治大 6-month holding period. Panel A. E-momentum strategies (12-month formation period with 3-month and 6-month holding period). 立. 3-month holding period. Return. CAPM Alpha. FF3 Alpha. 0.0083. 0.0066. 0.0061. 0.0098. 0.0143*. 0.0131. (1.17). (0.87). (0.80). (1.38). (2.02). (1.83). 0.0041. 0.0029. 0.0031. 0.0091. 0.0099*. (0.75). (0.49). (0.52). (1.88). (2.00). -0.0042. -0.0037. -0.003. -0.0007. -0.0044. (-1.15). (-1.15). (-1.00). (-1.02). (-1.77). (1.85). y. -0.0038. sit. (-1.66). er. io. al. 0.0093. ‧. Nat. P5E5-P1E1. FF3 Alpha. ‧ 國. P5E5. CAPM Alpha. 學. P1E1. Return. n. Panel B. E-momentum strategies (12-month formation period with 9-month and 12-month holding period) 9-month holding period P1E1 P5E5 P5E5-P1E1. Ch. i n U. v. 12-month holding period. eReturn n g c h iCAPM Alpha. Return. CAPM Alpha. FF3 Alpha. FF3 Alpha. 0.0102. 0.0137. 0.0142. 0.0104. 0.0085. 0.0101. (1.39). (1.83). (1.88). (1.33). (1.06). (1.26). 0.0115**. 0.0143**. 0.0141**. 0.0078*. 0.0073. 0.0076. (2.50). (3.13). (3.05). (1.97). (1.81). (1.88). 0.0013. 0.0006. -0.0001. -0.0026. -0.0012. -0.0025. (-1.06). (-1.36). (-1.47). (-1.33). (-1.27). (-1.43). 36. DOI:10.6814/NCCU201900078.

(43) Panel C. S-momentum strategies (12-month formation period with 3-month and 6-month holding period) 3-month holding period. P5S5 P5S5-P1S1. Return. CAPM Alpha. FF3 Alpha. Return. CAPM Alpha. FF3 Alpha. 0.0143. 0.0148. 0.0144. 0.0130. 0.0199**. 0.0190*. (1.73). (1.72). (1.68). (1.65). (2.60). (2.48). 0.0004. -0.0002. 0.0001. 0.0055. 0.0068. 0.0059. (0.09). (-0.03). (0.02). -0.0139. -0.0150. -0.0143. (-1.81). (-1.93). 立 (-1.81). (1.56) 政(1.29) 治大 -0.0075 -0.0131 (-1.11). (1.35) -0.0131. (-1.91). (-1.82). 學. ‧ 國. P1S1. 6-month holding period. Panel D. S-momentum strategies (12-month formation period with 9-month and 12-month holding period) Return. CAPM Alpha. FF3 Alpha. 0.0140. 0.0168*. 0.0165*. 0.0165*. 0.0124. 0.0136. (1.80). (2.12). (2.06). (1.98). (1.49). 0.0058. 0.0073. 0.0067. 0.0062. 0.0053. (1.25). (1.55). (1.38). (1.16). -0.0082. -0.0095. al. (-1.21). (-1.43). sit. er. (1.41). y. FF3 Alpha. iv -0.0098 -0.0071 n C h -0.0103 e(-1.48) (-1.39) n g c h i U(-1.15). n. P5S5-P1S1. CAPM Alpha. io. P5S5. Return. Nat. P1S1. 12-month holding period. ‧. 9-month holding period. (1.64) 0.0059 (1.29) -0.0077 (-1.30). 37. DOI:10.6814/NCCU201900078.

(44) Panel E. G-momentum strategies (12-month formation period with 3-month and 6-month holding period) 3-month holding period. P5G5 P5G5-P1G1. Return. CAPM Alpha. FF3 Alpha. Return. CAPM Alpha. FF3 Alpha. 0.0102. 0.0111. 0.0109. 0.0140*. 0.0191**. 0.0182**. (1.60). (1.67). (1.63). (2.09). (2.89). (2.75). 0.0054. 0.0057. 0.0056. 0.0094*. 0.0113*. 0.0103*. (1.14). (1.13). (1.09). -0.0048. -0.0054. -0.0053. (-1.06). (-1.25). 立 (-1.13). (2.49) 政(2.10) 治大 -0.0046 -0.0078 (-0.88). (2.24) -0.0079. (-1.63). (-1.59). 學. ‧ 國. P1G1. 6-month holding period. Panel F. G-momentum strategies (12-month formation period with 9-month and 12-month holding period) Return. CAPM Alpha. FF3 Alpha. 0.0156*. 0.0193**. 0.0188**. 0.0155*. 0.0124. 0.0132*. (2.35). (2.92). (2.83). (2.41). (1.90). 0.0096*. 0.0117*. 0.0114*. 0.0037. 0.0033. (1.98). (2.43). (0.77). (0.66). -0.006. -0.0076. al. (-1.36). (-1.86). sit. er. (2.31). y. FF3 Alpha. iv -0.0074 -0.0091 n C h -0.0118* e(-2.15) (-1.86) n g c h i U(-1.83). n. P5G5-P1G1. CAPM Alpha. io. P5G5. Return. Nat. P1G1. 12-month holding period. ‧. 9-month holding period. (2.03) 0.0039 (0.77) -0.0093* (-2.06). 38. DOI:10.6814/NCCU201900078.

(45) Table 7 Returns to small size ESG-momentum portfolios from 2010 to2017 This table reports the detailed excess returns of 25 groups we divided into, which are the pairs of 5 past-return groups and 5 ESG-Score groups with 12-month formation period, and their size are the smallest 10% during 2010 to 2017. The rightmost column and the bottommost column are the result of group subtraction. Panel A with 3-month holding period, Panel B with 6-month holding period, Panel C with 9-month holding period and Panel D with 12-month holding period. ***, **, and * indicate statistical significance at the 1%, 5%, and 10 % level, respectively.. 政治大 ESG4 ESG5(High). Panel A. 12-month formation period with 3-month holding period. 立. ESG3. 0.0203**. 0.0207*. 0.0205*. 0.0127. 0.0614. 0.0411. P2. 0.0130. 0.0072. 0.0327**. 0.0184. 0.0337. 0.0207. P3. 0.0163*. 0.0021. 0.0026. 0.0331. 0.0288. 0.0125. P4. 0.0065. 0.0267*. 0.0186*. 0.0344*. -0.0601. -0.0666. P5(Winner). 0.0117. 0.0095. 0.0050. 0.0182. -0.0539. -0.0086. -0.0112. -0.0155. 0.0055. -0.1153. ESG4. ESG5(High). 0.0178. 0.1035. v. 0.0954. -0.1132. -0.1262. ‧. -0.0656. y. Nat. -0.0742. sit. P1(Loser). ‧ 國. ESG2. 學. ESG1(Low). al. ESG1(Low). ESG2. P1(Loser). 0.0081. 0.0167. 0.0101. P2. 0.0130. 0.0147*. 0.0199*. P3. 0.0019. 0.0183*. 0.0057. 0.0207. -0.1104. -0.1123. P4. 0.0157*. 0.0106. 0.0054. 0.0547*. 0.0874**. 0.0717**. P5(Winner). 0.0184*. 0.0147. 0.0038. 0.0100. -0.0509. -0.0693. 0.0103. -0.0020. -0.0063. -0.0078. -0.1544. -0.0590. n. ESG3. er. io. Panel B. 12-month formation period with 6-month holding period. Ch. e-0.0005 ngchi. i n U. 39. DOI:10.6814/NCCU201900078.

(46) Panel C. 12-month formation period with 9-month holding period ESG1(Low). ESG2. ESG3. ESG4. ESG5(High). P1(Loser). 0.0059. 0.0070. 0.0038. 0.0272. 0.1157. 0.1098. P2. 0.0113. 0.0199*. 0.0092. 0.0109. 0.0091. -0.0021. P3. 0.0107. 0.0152. 0.0394. 0.0127. -0.0858. -0.0965. P4. 0.0102. 0.0216*. 0.0153. 0.0142. 0.0986. 0.0884. P5(Winner). 0.0122. 0.0143. 0.0265. 0.0063. 0.0073. 0.0227. 立. 政0.0038治 -0.1314** 大 -0.0234 -0.2471. P3. 0.0158*. 0.0036. P4. 0.0175*. 0.0143. P5(Winner). 0.0154*. 0.0125. 0.0031. -0.0045. 0.1587. -0.0048. 0.0511. -0.1158. 0.0006. 0.0048. -0.0276. 0.0031. -0.0107. -0.0997. 0.0360**. 0.0056. -0.1037. -0.0318. -0.2624. al. 0.0185. Ch. n engchi U. iv. 0.1464 -0.1182 -0.0434. y. 0.0099. 0.0374. sit. 0.0024. 0.0175. er. P2. ‧ 國. 0.0170. ESG5(High). ‧. 0.0123. ESG4. n. P1(Loser). ESG3. io. ESG2. Nat. ESG1(Low). -0.1373*. 學. Panel D. 12-month formation period with 12-month holding period. -0.1436**. -0.1172 -0.1191* -0.1160*. 40. DOI:10.6814/NCCU201900078.

(47) Figure 1 Distribution of ESG-Scores This figure reports the distribution of ESG-Scores, E-Scores, S-Scores and G-Scores separately by year, and it’s more easy for us to observe the trend of each scores. We can also compare them from Figure1.5. Data period is from 2005 to 2017.. 25. 25. 立. ‧ 國. 15. 2010. Figure 1.1 Distribution of ESG-Scores. 2012. 2014. al. 2016. n. 2008. sit. io. 2006. y. 5. Nat. 5. 10. ‧. 10. 0 2004. 20. 學. 15. 政治大. 0 2004. 2018. Ch. er. 20. v 2008 i n. 2006. 2010. 2012. 2014. 2016. 2018. U of E-Scores e nFigure g c h1.2i Distribution. 41. DOI:10.6814/NCCU201900078.

(48) 25. 55 54.5. 20. 54 53.5. 15. 53. 政治大 52.5. 立. 2014. 2016. 50 2004. 2018. Nat. Figure 1.3 Distribution of S-Scores. 2012. 2006. 2008. 2010. 2012. 2014. 2016. 2018. y. 2010. 50.5. Figure 1.4 Distribution of G-Scores. io. sit. 2008. 51. n. al. er. 2006. 51.5. ‧. 0 2004. 52. 學. 5. ‧ 國. 10. Ch. engchi. i n U. v. 42. DOI:10.6814/NCCU201900078.

(49) 60 50 40. 政治大. 30. 立. ESG-Score. 2008. 2010 E-Score. 2012. 2014. 2016. S-Score. 2018. G-Score. Nat. y. 2006. ‧. 0 2004. 學. 10. ‧ 國. 20. n. er. io. al. sit. Figure 1.5 Distribution of ESG-Scores, E-Scores, S-Scores and G-Scores. Ch. engchi. i n U. v. 43. DOI:10.6814/NCCU201900078.

(50)