國立交通大學 財務金融研究所 碩士論文 股利政策對現金增資公司的長期績效影響 The Impact of Dividend Policy on the Long Term Performance of SEOs Firms 研究生 : 黃渝薇 指導教授 : 林建榮博士 中華民國九十七年六月

國立交通大學 財務金融研究所 碩士論文 股利政策對現金增資公司的長期績效影響 The Impact of Dividend Policy on the Long Term Performance of SEOs Firms 研究生 : 黃渝薇 指導教授 : 林建榮博士 中華民國九十七年六月

股利政策對現金增資公司的長期績效影響 The Impact of Dividend Policy on the Long Term Performance of SEOs Firms 研究生 : 黃渝薇 指導教授 : 林建榮博士 Student : Yu-Wei Huang Advisor : Dr. Jane-Raung Lin 國立交通大學 財務金融研究所碩士班 碩士論文 A Thesis Submitted to Graduate Institute of Finance National Chiao Tung University in Partial Fulfillment of the Requirements for the Degree of Master of Science in Finance June 2008 Hsinchu, Taiwan, Republic of China 中華民國九十七年六月

股利政策對現金增資公司的長期績效影響 研究生 : 黃渝薇 指導教授 : 林建榮博士 國立交通大學財務金融研究所碩士班 2008 年 6 月 摘要 : 學者已證實現金增資公司會有長達三年以上的負報酬現象, 而股利增加 ( 減少 ) 也會使得長期報酬有正向 ( 負向 ) 反應 本研究旨在討論股利對現金增資公司的長期績效影響, 因此以現金增資公司的股利政策來做分類, 發現有發放股利的公司在進行現金增資時, 會有較優異的報酬, 次為沒發放股利的公司, 報酬表現最差的公司為在現金增資後停止發放股利的公司, 並且持續發放越久股利的公司會有越好的報酬表現 此結果符合資訊不對稱與股利訊號理論的結論, 即現金增資公司會有負長期報酬的效果, 但可藉由股利政策來降低負報酬 而有發放股利的公司通常有較高的帳面對股東權益價值與資產價值, 並且其公司營運期間也較久, 顯示有發放股利的現金增資公司是較為穩定及成熟的公司 關鍵字 : 現金增資 ; 資訊不對稱 ; 長期績效 ; 股利 i

The Impact of Dividend Policy on the Long Term Performance of SEOs Firms Student: Yu-Wei Huang Advisor: Dr. Jane-Raung Lin Graduate Institute of Finance National Chiao Tung University June 2008 Abstract Many scholars indicate that firms after SEOs announcement would sustain long term negative performance for more than three years. However, if firms increase (decreasing) its dividend payout would result in positive (negative) return in the long run. This paper examines the impact of dividend policy on the long term performance of SEOs firms. Categorizing by dividend policy, if the firms keep paying dividend after SEOs announcement perform best in long term performance, non-dividend-paying firms get the second best result, and the worst is firms paying dividend around SEOs and cut off after SEOs announcement. Moreover, firms continue paying dividend for longer period before SEOs would get less long term negative returns. This result is consistent with asymmetric information and signal theory. The firms paying dividend statistically have higher ROA, market value, total asset, and firm age, which means that most of dividend-paying firms are larger, more stable, and more mature. Keywords: Seasoned Equity Offering; asymmetric information; Long Term Performance; Dividend ii

誌 謝 這本論文的完成, 最需要感謝的就是我的指導教授林建榮老師, 感謝老師一步步的指導我論文的方向, 細心的修正我的論文, 讓我了解到做學問的方法, 並且順利將論文完成, 拿到碩士學位 兩年的時間真的過得很快, 從一開始來新竹有些許的不情願, 到現在真的要離開新竹, 不能再與財金所的同學們朝夕相處時的感傷, 兩年間所發生的事歷歷在目, 在寫論文這段苦悶的日子裡, 感謝那些幫助過我與陪我歡笑的學長姐與同學們 首先感謝跟我同為林老師指導的同學們, 在撰寫論文過程中一起討論問題, 事半功倍, 尤其是我的室友詩政, 常常都一路奮戰到熄燈 當我心情不好或是功課方面有問題時, 感謝以文都會不厭其煩的幫忙解決, 還會常常說一些好笑的事跟大家分享, 還有感謝文娟常常陪我們在宿舍聊天, 解決數學方面的難題 謝謝本班的好人好事代表建佑與文誠常常帶我出去吃飯, 讓我不用天天吃二餐 還有熱心助人的小田, 謝謝你常常搞笑, 還幫忙印初稿, 二話不說就願意幫我處理論文的數據 演什麼像什麼的阿 Sam 國台英語都超流利又很會唱歌的俊儒, 看到你們在那邊搞笑就會很開心 還有感謝跟我玩過撲克牌跟麻將的各位, 讓我學到很多賺大錢的技巧 感謝在新竹認識的每一位同學, 有你們讓我在這兩年過的多采多姿, 從招生說明會到謝師宴, 都是美好的回憶 還有要感謝正在保家衛國的文欽, 雖然你不懂我的論文題目在做什麼, 但是你總是會努力的想辦法幫我解決問題, 當我心煩意亂心情不好時, 也願意包容我的情緒, 讓我無後顧之憂 最後要感謝我的爸媽 妹妹渝慧, 還有在天上的咪咪, 你們都是最支持我的後盾, 每當我在論文上遇到不順時, 幸虧有你們的鼓勵與打氣, 讓我有繼續往前邁進的勇氣, 謹將這本論文獻給我最親愛的家人 師長 同學 與所有關心我的朋友們 黃渝薇誌於國立交通大學財務金融所碩士班 中華民國九十七年六月於新竹 iii

Content 1. Introduction... 1 2. Literature Review... 4 2.1 Dividend... 4 2.2 Seasoned EquityOofferings... 5 3. Data and Descriptive Statistics... 9 3.1 Data Selection... 9 3.2 Descriptive Statistics and Sample Characteristics... 10 4. Methodology... 11 4.1 Buy-and-Hold Abnormal Returns (BHARs)... 11 4.2 Calendar-time Fama and French Three-Factor Model Portfolio Regressions... 13 4.3 Multiple Regression... 14 5. Empirical Result... 16 5.1 Long Term Performance of SEOs Firms... 16 5.2 Long Term Performance Categorized by Dividend-Paying Firms or not... 16 5.3 Long Term performance Categorized by Keeping Paying Dividend or not and Regression... 17 5.4 Long Term Performance Categorized of Dividend-Paying Firms by Different Definition... 20 6. Conclusion... 22 Reference... 24 iv

List of Tables Table I Distribution of Seasoned Equity Offerings by Year and Industry... 27 Table II Variable Characteristic of SEOs Firm... 29 Table III Five Year Buy-and-Hold Abnormal Returns for All SEOs Firms... 31 Table IV Calendar-Time Fama and French Three-Factor Model Portfolio Regression for All Firms... 32 Table V Five Year Buy-and-Hold Abnormal Returns for Dividend-Paying Firms and Non-Dividend-Paying Firms... 33 Table VI Calendar-Time Fama and French Three-Factor Model Portfolio Regression for Dividend-Paying Firms and Non-Dividend-Paying Firms... 34 Table VII Five Year Buy-and-Hold Abnormal Returns for Keeping-Dividend-Paying Firms and non-keeping-dividend-paying Firms after SEOs... 35 Table VIII Calendar-Time Fama and French Three-Factor Model Portfolio Regression for Keeping-Dividend-Paying Firms and Non-Keeping-Dividend-Paying Firms after SEOs... 37 Table IX Ordinary Least Squares Regressions of Abnormal Stock Returns to Keeping-Dividend-Paying Firms and Non-Keeping-Dividend-Paying Firms after SEOs... 38 Table X Five Year Buy-and-Hold Abnormal Returns for Different Definition of Dividend-Paying Firms... 39 v

1. Introduction Many empirical researches document that when announcing a new equity issue, firms would have a significant decline in stock returns around the announcement day. Asquith and Mullins (1986) show that when firms declared seasoned equity offerings (SEOs), firms stock price would decline by 2-3%. Loughran and Ritter (1995) calculate the difference in returns between issuers and non-issuers, and find that an investor would have to invest about 44% more money in the issuers than in the non-issuers to get the same rewards after five year. How do we explain the phenomenon? Myers and Majiluf (1984) attribute it to the asymmetric information between managers and investors. When a firm s stock price is overpriced, the manager would be likely to announce equity offering. When investors found the fact, they would treat SEOs as a bad signal and revise the stock price downward. If managers reduce asymmetric information, the SEOs negative return might decline. To reduce the asymmetric information has many kinds of methods. One way is to change dividend policy. Lintner (1956) proposes that managers change dividend by their anticipative profits of firm. Miller and Modigliani s (1961) irrelevance theorems show that in the prefect market, a company s dividend policy does not affect it s value. John and Williams (1985), Miller and Rock (1985) show that declaring dividend is considered a good signal, which are costly for bad firms to mimic. Michaely, Thaler, and Womack (1995) prove that dividend initiation and omission would influence not only the announcement return but the long-run abnormal return. Loderer and Mauer (1992) investigate whether managers announce dividends before SEOs to decrease negative returns and whether the action can coordinate the stock price. Their result doesn t support the assumptions, but managers appear 1

reluctant to cut dividends before the SEOs announcement. That is to say, what dividend releases to investors is important to managers. Chang Bin (2006) mentioned that after 1985, the market reacts less negative return to a dividend payer s SEO announcement than to non-dividend payer. Fama and French (2001) found that the proportion of firms paying cash dividends falls from 66.5% in 1978 to 20.8% in 1999. In this article, we try to examine whether dividend policy has impact on the long term performance of SEO firms. We collect the sample of the SEOs firms with IPOs after 1970. According to firms dividend policy, dividing SEOs firms into two groups: dividend-paying firms and non-dividend-paying firms. We observe that the dividend-paying firms have longer operating year, larger market value, and higher book to market ratio than non-dividend-paying firms. This means that dividend-paying firms are stable and mature than non-dividend-paying firms. Next, We further divide the dividend-paying firms into two groups depend on whether they keep paying dividend after SEOs. We find that firms paying dividend both before and after SEOs announcement have the best performance, and firms paying dividend before SEOs announcement but cutting after SEOs have the worst performance. We use the buy-and-hold abnormal return (BHAR) and Fama and French three factor model (1993) to estimate the abnormal return. we change the definition to categorize dividend-paying firms. The definition of dividend-paying firms is based on how long a firm pay cash dividend before the stock-offering date. We find that firms paying longer dividend before SEOs announcement would get better long term abnormal return. The result shows that the steadier dividend policy, the less negative return after seasoned equity offerings. The remainder of the paper is organized as follows. Section 2 is a review of previous literature, addressing both theoretical and empirical aspects in the return of SEOs and the asymmetric information between IPO, SEOs, and dividend. In Section 3, 2

we discuss the sample selection and descriptive statistics. We introduce the methodology of evaluating long-run performance in section 4. Section 5 presents the empirical results about the long run performance of SEOs. Finally, conclusions are provided in the last section. 3

2. Literature Review 2.1 Dividend 2.1.1. Asymmetric Information and Dividend Signal Lintner (1956) notes that how many dividends manager want to pay depends on the company s future profitability in the long run and manager would prefer to pay dividend smoothly and steadily. That is to say, if manager pays dividend at the first time or increases dividends, the market will treat it as a good signal. Miller and Modigliani s (1958, 1961) irrelevance theorems show that in the prefect market, company s dividend policy don t affect company s value because managers and investors have symmetric information. Without prefect market, Miller and Rock (1985) think that dividends didn t affect company s value, but dividends would release signals to let investors know the firm s condition. Easterbrook (1984) finds that the stockholders need to bear monitor cost. After paying dividends, managers may borrow money from banks or lenders. There would be more market power to monitor these firms. Jensen (1986) mentions that the relationship between free cash flow and agency problem. The higher free cash flow would cause the bigger agency problem. So that firms pay dividend to stockholder would lower free cash flow. From the point of views, we can say that paying dividends is good news to investors. Healy and Palepu (1988) find that when firms announce to pay (omit) dividends, the profits of firms would increase (decrease) obviously in the following year, so do the stock price. But if firms only have the announcement of profits, the stock market wouldn t have such big fluctuation. Denis, Denis, and Sarin (1994) examine the cash flow signaling, overinvestment, and dividend clientele explanations for the information content of dividend change announcement. They 4

found the announcement returns are positively change to dividend change, dividend yield. Those findings support that cash flow signaling and dividend clientele hypotheses. 2.1.2The Long-Run Performance of Dividends Different dividend policy would influence the firm s stock return, but how does dividend influence return in the long run? There are three possible aspects. First, to treat dividend as an earning announcement. Bernard and Thomas (1990) told that when firms make surprising earnings announcements, the return would move in the same direction for the next few quarts. Second, some literatures make a description of overreaction or reversion in price, and the long-run return would go to the opposite direction. Third, changing dividend may cause a change in the type of stockholders. This is what we call clientele effect. Bajaj and Vijh (1990) suggested that the existence of dividend clienteles may partially explain price reactions to dividend change. Michaely, Thaler, and Womack (1995) used the data during 1964 to 1988. They find after the announcement of dividend change, price continue to drift in the same direction. They use equally-weighted market index as benchmark, and they proofed the stock price continue to change in the same direction even after the announcement over three years. 2.2 Seasoned equity offerings 2.2.1 Asymmetric Information of Seasoned Equity Offering By pecking order theory, when managers do financing decisions, they would use internal financing first, and SEOs are the last resort. Asquith and Mullins (1986) prove that when announcing SEOs, the stock returns decline by 2-3%. Myers and 5

Majluf (1984) explain that it is because of the asymmetric information between managers and stockholders. They contended only if firms equities are overpriced, managers would issue equities. Consequently, rational investors anticipate this behavior, and they would discount the price of issuing firms. Myers and Majluf (1984) proposed if managers can decrease the asymmetric information, and let investors know the firm s demand of SEOs. The negative return around the announcement day wouldn t so serious. Korajczyk, Lucas, and McDonald (1992) show that firms time equity issues after some information releases to decrease valuation uncertainty, and the negative return would decrease. The information announcement effect of the return is negative related to time internal. That is to say, the longer time internal between information releases and announcement of offerings, lead to the less effect on return. D Mello, Tawatnuntachai, and Yaman (2003) found that firms conduct multiple equity issues have less negative announcement return. After conducting more than twice SEOs, the abnormal return isn t different from zero. Bayless and Chaplinsky (1996) think that without considering the specific of firms, the whole market may affect the abnormal announcement return of SEOs. The negative returns of firms which offer equities in high volume of circulation market are less than firms which offer equity in low volume of circulation market about 200 basis points. Brous (1990) researched that whether common stock offering announcements convey information about the level of the firm s future cash flows. The forecasts of the current year earnings are, on average, decreased when firms announce plans to issue additional common stock. The size of the decrease is significantly related to announcement period abnormal stock returns. In contrast, forecasts of the five-year growth rate of earnings are, on average, unchanged. He thinks equity offering 6

announcement conveys unfavorable information. What influence would be if we combine with the research of dividend signal and SEOs? Loderer and Mauer (1992) investigate whether managers announce dividends before SEOs to decrease the effect and whether the action can coordinate the stock price. The evidence doesn t support the assumptions, but managers appear reluctant to cut dividends before the SEOs announcement. 2.2.2 The Long-Run Performance of SEOs Loughran and Ritter (1995) show that the IPOs and SEOs firms during 1970 to 1990 underperform relative to nonissuing firms for five years after the offering date significantly. During the five years after the SEOs, the investors have received average return only 7% per year. There are two possible reasons to explain the phenomenon. First, they found the degree to which issuing firms underperform varies over time. When the offering year with little issuing activity, the issuing firm face slight underperformance. On the other hand, firms selling stock during high-volume periods severely underperform. Secondly, they showed that the issuing firms risk, beta, is higher than nonissuing firms. That implying issuing firms should have higher return. Katherine and Affleck (1995) use the data during 1975-1989 to check the underperformance of SEOs firm. They use the matched firms from the same industry and similar size that did not issue equity. Their control variable are issuing year, firm s age, book to market ratio, firm s size, exchange. They found that underperformance is existence with every subgroup, but the most significant severe for the smallest, youngest, lowest book to market ratio, and Nasdaq-traded firms. With those evidences like Ritter s (1990) conclusion with IPOs firm, they concluded that manager is able to take advantage of firm-specific information to issue equity when the firm s stock is overvalued. 7

Eckbo, Masulis, and Norli (2000) try to adjust risk to find matching firms. They noted that that firms issuing stock makes a little higher market risk than nonissuing firms, but that the less specific risk would arise such as unanticipated inflation, default risk. SEOs firms have lower leverage ratio which means they have less unanticipated inflation and default risk, and they have higher stock liquidity which the nonissuing firms don t change. They thought the reason why issuing firm have lower abnormal return is because of risk-unadjusted. 8

3. Data and Descriptive Statistics 3.1 Data Selection The data include seasoned equity offerings (SEOs) by US companies from January 1985 to December 2002, but excludes equity offerings by closed-end funds, real estate investment trusts (REITs), unit investment trusts, and American depositary receipt (ADRs); in other words, the data s share code for 10 or 11. IPO firms after 1970 with seasoned equity offering during1985 2002 are collected from Securities Data Corporation (SDC). In addition, the sample must meet the following criteria: 1. The issuing firms are not utilities (with first two digit SIC code of 49) or financial institutions (with first digit SIC code of 6). 2. The issuing firms have monthly returns which can be obtained from the Center for Research in Security Prices (CRSP). To calculate the long term abnormal returns, we need monthly return at least one year after SEOs. 3. The issuing firm must have enough financial information for analysis from Compustat. 4. To reduce dependence for the statistical tests, we follow Healy and Palepu's (1990) procedure and exclude SEOs by the same firm during the following five years after SEOs in our sample. In this article, we select the sample of the SEOs firms which had their IPO after 1970, and we observe the firms dividend policy to divide into two groups: dividend-paying firms and non-dividend-paying firms. To define what firms are dividend payers or non-dividend payers, we follow Loderer and Mauer s (1992) definition. Dividend-paying firms are defined as firms that pay at least one cash dividend during the three quarters preceding or following the SEOs date, and the 9

others are non-dividend-paying firms. And then, we divide the dividend-paying firms into two groups based on whether they keep paying dividend after SEOs or not: keeping-dividend-paying firms and non-keeping-dividend-paying firms. 3.2 Descriptive Statistics and Sample Characteristics According to above definition, the full sample contains 1333 SEOs events issuing in the NYSE, AMEX, NASDAQ market from January 1987 to December 2002. Table I is the distribution of SEOs by year and industry. Panel A shows the time distribution of the final sample. From Panel A, we can find that the number of firms increase in 1987, and drop off seriously form 1999. Panel B exhibits the SIC distribution of the SEOs firm. The information in Panel B shows that the SEOs firms are concentration of the Computer Hardware & software, Electric and electronic equipment. The industry of Computer Hardware & software are made 265 of the 1333 total sample offers (19.94%). Table II display the descriptive statistics of SEOs firm for all sample, dividend-paying firms, and non-dividend-paying firms. From Table II, we can detect the different characteristic between dividend-paying firms and non-dividend-paying firms. In panel B, the dividend payers have higher ROA, market value of equity, total asset, book to market ratio, and firm age 1, which means relative to non-dividend payers, most of dividend-paying firms are large, stable, and mature. 1 We use pooled T or satterhwaite T to test the difference of mean. We find that DLTT/AT, ROA, MVA, BMK, DAT, AT, logmva, and AGE are significantly different from dividend-paying firm and non-dividend-paying firms. 10

4. Methodology Kothari & Warner (1997), Fama (1998), Loughran & Ritter (2000), Brav (2000), and Mitchell & Stafford (2000) argue that different methodology will affect the long-run abnormal performance. In this paper, we use the buy-and-hold abnormal return method (BHARs) and Fama and French s three factor model (1993) to evaluate robustness of SEOs post performance. 4.1 Buy-and-Hold Abnormal Returns (BHARs) Barber and Lyon (1997) and Lyon et al. (1999) argue that BHARs are important because they precisely measure investor experience from buying and holding securities for a period. However, they also found the common estimated method may bias the estimates. This bias result from the new listing, rebalancing of benchmark portfolio, and skewness of multiyear. In order to solve the skewness problem, we follow Barber and Lyon s (1999) bootstrapped skewness-adjusted t-statistic. For each firm i, BHR is a investor s holding return from the announcement month to time T, calculating as follows: 1 BHR it R it T t 1 (1) Where R it is the monthly return of firm i The buy-and-hold abnormal return (BHAR) is defined as follows, BHAR it T 1 Rit 1 E( Rit ) T t 1 t 1 (2) Where E R ) is the benchmark return. In this paper, we use three different ( it benchmarks: (1) the CRSP equally-weighted market portfolio, (2) the CRSP value-weighted market portfolio, (3) a size and book-to-market matched control sample. 11

To find the size and book-to-market match control firm, we follow Barber and Lyon s (1999) criterions. First, we identify all the firms in the CRSP database with the market value of common equity between 70% and 130% of the market value of equity of a sample firm. Second, from the size set of firms, we choose the firm with the closest book-to-market ratio to that of the sample firm. Market value of common equity measured on the first day of the issue month. Book-to-market ratio is the firm s book value of equity divided by its market value of equity, measured at the fiscal year end prior to the issue. And the match firm is in the NYSE, AMEX, and NASDAQ market around 1987 to 2002 and we excludes firms involved in IPO or SEO events five years prior. We use two t-statistics to test the null hypothesis that the mean buy-and-hold abnormal return is equal to zero for a sample of n firms. We first display a conventional t-statistic: t AR ( AR T T ) / n (3) Where AR is the sample mean and AR ) is the cross-sectional sample standard T ( T deviation of abnormal returns for the sample firm. The other is the bootstrapped skewness-adjust t-statistic, advocated by Barber and Lyon s (1999): t 1 2 n S ˆ S 3 1 ˆ 6n (4) Where S Mean( BHAR) ( BHAR) t t, and ˆ n i 1 AR it n ( AR AR T ) 3 T 3 Note that γ is the estimation of the skewness coefficient. Barber and Lyon (1999) 12

said that the bootstrapped application of the t-statistic should be the better statistic to the t test when the distribution is asymmetrical. According to Barber and Lyon (1999), the distribution of BHARs is positively skewed and generally doesn t have a zero mean. That why we use the skewness-adjust t-statistic to test the mean. Additionally, the BHAR methodology assumes all observations are independent of one another. Therefore, we use calendar-time Fama and French three-factor model portfolio regressions to fix the problem. 4.2 Calendar-time Fama and French Three-Factor Model Portfolio Regressions Fama and French (1993) mentioned that firms market factor, size, and book to market ratio have strong relationship to firms return. If we want to know the performance of an event portfolio, we can construct an event portfolio in calendar time. The calendar-time portfolio approach was suggested by Fama (1998) and Mitchell and Stafford (2000) which represents an improvement over the BHAR methodology. The equation is as follow: R it R ( R R ) s SMB ft i i mt ft i i h HML i i it (5) where R it is the return on the portfolio i in month t, R ft is the return on one-month Treasury bills in month t, R mt is the return on a value-weighted market index in month t, SMB t is the difference in the returns of a portfolio of the small and big stocks in months t, HML t is the difference in the returns of a portfolio of high book-to-market stocks and low book-to-market stocks in the month t. The intercept, i, means the average monthly abnormal return in the holding period. 13

To construct calendar portfolio, we form an equal-weighted and a value-weighted portfolios each month of all sample firms that participated in the event within the previous 5 years 2. The calendar time portfolios would rebalance monthly to drop all companies that reach the end of their 5-year period and add all companies that have just executed a transaction. The methodology s feature is to form a portfolio by calendar month, and the cross-sectional dependence problem which occurs in BHAR methodology is lower. But in order to form the calendar portfolio, the test power is sacrificed. 4.3 Multiple Regression The sample consists of 1333 SEOs events which occur during the period between January 1987 and December 2002 that are selected from Securities Data Company s New Issues Database. We find several variables and try to observe the relationship between the abnormal return and firm characteristics. The multiple regression model is described as follows: BHAR ( AGE) ( DAT) ( BMK) ( CAPX 0 (log MVA) ( KDIV ) 6 1 2 7 3 4 / AT ) ( ROA) 5 (6) The dependent variable is the buy-and-hold abnormal return (BHAR). BHAR is the average monthly abnormal returns estimated by the benchmarks which of CRSP equally-weighted market portfolio. KDIV is a 0/1 indicator variable equal to 1 for keeping-dividend-paying firms and 0 otherwise. 2 We exclude multiple observations on the same firm that occur within 5 years of the initial observation. 14

We place some firm characteristics as the control variables, where AGE is the issuer s operating years; DAT is total debt to total asset; BMK is book value of equity to market value of equity; CAPX/AT is total capital expenditure to total asset; ROA is return on asset; logmva is log market value of asset, where MVA equals total asset minus total book value of equity pluses total market value of equity. 15

5. Empirical Result 5.1 Long Term Performance of SEOs Firms Table III presents the long-term abnormal performance for five years after issuing SEOs during January 1987 to December 2002. Returns are calculated by buy-and-hold abnormal returns with three different benchmarks. The size and B/M ratio matching firms return pattern is similar to Loughran and Ritter (1995). At first year, the SEO-issuers have slight negative return and then underperform thereafter. The CRSP equally-weighted portfolio and the CRSP value-weighted portfolio are similar result in abnormal return. In Table IV, we use Fama and French three-factor model to test the abnormal return. We find the underperformance is insignificant negative except the second year (-0.4%). 3 5.2 Long Term Performance Categorized by Dividend-paying firms or not In this section, we divide our sample into two groups. We classify firms into dividend-paying and non-dividend-paying firms. We follow Loderer and Mauer s (1992) to define dividend payers and non-dividend payers. Dividend-paying firms are defined as firms that pay cash dividend during the three quarters preceding or following the stock-offering date, and the others are non-dividend-paying firms. Table V reports the average buy-and-hold abnormal month return for five year after SEOs for dividend-paying firms and non-dividend-paying firms. Based on the size and B/M ratio benchmark, we compare the difference between dividend-paying firms and non-dividend-paying firms. At the first year after issuing, both Panel A and 3 The dependent variable is the equally-weighted month portfolio return. We use equally-weight and value- weight month portfolio return, and the return pattern is similar. 16

Panel B have slight negative abnormal return. we find the negative abnormal return is insignificant during issuing after five years in Panel A range from -4.12% to 4.6%. On the contrary, Panel B shows the significant negative abnormal return range from -0.19% to -13.07%. Under the CRSP equally-weighted portfolio and CRSP value-weighted portfolio, both Panel A and Panel B have similar return pattern that dividend-paying firms have less negative return. The negative return is more significantly in CRSP equally-weighted portfolio, and the phenomenon is more obvious in Panel A. This result is similar to Loughran & Ritter (1995) and Mitchell & Stafford (2000). The result reflect that the pattern that smaller offerings underperformance more than larger offerings, and the small issues can t drive the abnormal return in the CRSP value-weighted portfolio. Table VI reports the Fama and French three-factor model. We can find that non-dividend-paying firms have significant negative abnormal returns (-0.49%) at the 5% level in the second year after issue offerings. In other years, both the dividend-paying firms and the non-dividend-paying firms have insignificant abnormal returns. In this classification, we use BHAR methodology and Fama and French three-factor model. We find that the result of BHAR is significant, but the result is consistent with the three factor model. That what we mention before the test power would be decreased by forming the calendar month portfolio. 5.3 Long Term Performance Categorized by Keeping Paying Dividend or Not and Regression In this section, we focus on the dividend-paying firms. We further divide those firms into two groups: keeping-dividend-paying firms define as firms keeping paying 17

dividend at least one year and non-keeping-dividend-paying firms. 5.3.1 Long Term Performance Table VII shows the result of buy-and-hold abnormal returns for keeping-dividend-paying firms and non-keeping-dividend-paying firms after SEOs. Panel A shows insignificant abnormal return of keeping-dividend-paying firms. Based on the three different benchmarks, there are positive abnormal returns in the first two year. The worst negative return displays mean abnormal return -18.56% at a quite insignificant level under a CRSP equally-weighted index in the fifth year. On the other hand, non-keeping-dividend-paying firms display mean abnormal return -92.07% at 1% significant level in the same period which show the extreme different abnormal return relative to keeping-dividend-paying firms. In panel B, except the first year after issuing, every mean abnormal return is negative and significant. Same as previous section, the CRSP equally-weighted portfolio has higher negative abnormal return than the CRSP value-weighted portfolio. It means that the small issue offerings have higher negative abnormal return. Table VIII is the Fama and French three-factor model (1993) for keeping-dividend-paying firms and non-keeping-dividend-paying firms after SEOs announcement. The keeping-dividend-paying firms have slight positive return in the first three year and worse in the fourth and fifth year. The t-statistic of keeping-dividend-paying firm is not significant during the five year after SEOs announcement. The non-keeping-dividend-paying firms have significantly negative abnormal return in the following five year after equity issue. We compare the difference between keeping-dividend-paying firms, non-keeping-dividend-paying firms, and non-dividend-paying firms (Panel B of table V). Using CRSP equally-weighted portfolio as benchmark, we find that the keeping-dividend-paying firms have the highest abnormal return ranged from 2.91% to -18.56%. The 18

non-dividend-paying firms is in the middle ranged from -3.45% to -31.86%. The non-keeping-dividend-paying firms are the worst abnormal return ranged from -6.01% to -92.07%. By Loderer and Mauer s (1992), the managers have the motive not to stop paying dividend before SEOs because investors treat cutting dividend as a bad signal. Loderer and Mauer s (1992) find that relative to non-dividend-paying firm, the dividend-paying firms don t have better abnormal return around the announcement date. Myers and Majiluf (1984) attribute the SEOs negative return arising from the asymmetric information between manages and investors, and paying dividend give out good signal. The non-keeping-dividend-paying firms pay dividend before the SEOs announcement and try to imitate good firms and lower the asymmetric information to get better return, and then cut it off. In the long run, the investors become aware of the fact. That is why the non-keeping-dividend-paying firms would have the worst abnormal returns. The result is also identical with Michaely, Thaler, and Womack s (1995) conclusion that the dividend change would make stock price continue to change in the same direction even after the announcement over three years. 5.3.2 Ordinary Least Squares Regression Results Table IX shows the ordinary least squares regression of post-issue stock return in the issue year for various holding periods. The dependent variable is the equally-weighted portfolio BHARs. KDIV is a dummy variable which equal to 1 for keeping-dividend-paying firms. We use the firm s operating year (AGE), total debt to total asset (DAT), book to market ratio (BMK), capital expenditure to total asset (CAPX/AT), return on asset (ROA), and logged market value of asset (logmva) as the control variables. The result in Table IX indicates the coefficients of KDIV are all positive, with 19

significantly positive coefficients at 5% or better for two-, three-, four-,and five-year holding periods. With the time goes by, the non-keeping-dividend-paying firms increase the weight of negative return. We also can detect that market value of asset has a negative relationship with long term abnormal return, and the book to market value has a positive relationship with long term abnormal return. Our overall evidence in Table IX suggests that keeping-dividend-paying firms perform significantly better the non-keeping-dividend-paying firms, consistent with the results in Table VII. 5.4 Long Term Performance Categorized of Dividend-Paying Firms by Different Definition In section 5.2, dividend-paying firms are the firms which pay cash dividend during the three quarters preceding or following the stock-offering date. In this section, we use different definition to category dividend-paying firms. The definition of dividend-paying firms is based on how long a firm pay cash dividend before the stock-offering date. Table X is the result of long term performance by different definition. Panel A is the same definition as section 5.2. In Panel A, the BHARs range from 0.9% to -28.36% for under the CRSP equally-weighted portfolio, and in the last two years BHARs reach the significant standard. Panel B is the definition of paying dividend for two years, and the BHARs range from 1.31% to -25.39% under the CRSP equally-weighted portfolio. Panel C is the definition of paying dividend for three years, and the BHARs range from 1.13% to -22.31% under the CRSP equally-weighted portfolio. We find that from table X, the negative returns get worse for longer time. The phenomenon is same as different definition, but the negative returns is slight in Panel B and Panel C, especially in Panel C. In Panel C, the 20

negative returns even do not reach the significant standard. The empirical result shows that the steadier dividend policy, the less negative return after seasoned equity offerings. 21

6. Conclusion In the study, we examine the long term performance of SEOs firms with regard to different dividend policy. Many previous studies show the long term return of seasoned equity offerings issuers and dividend payers or not. They found that issuing SEOs and cutting dividend would lead to the significant long term negative returns 4. But they seldom combine the effect of SEOs and dividend policy to see the long term return. We use 1331 SEOs issuing during from 1987 to 2002 collected from Securities Data Corporation to see the outcome. The main conclusions are as follow: 1. The Dividend-paying firms have larger firm size, higher return on asset, higher book to market ratio, and longer operating year. 2. The Dividend-paying firms have less long term negative returns than the Non-dividend-paying firms. 3. The firms which continue paying dividend after SEOs announcement would have the best long term performance regardless of using BHARs or calendar-time Fama and French three-factor model portfolio regressions. The second is the Non-dividend-paying firms. The worst is the firm which pay dividend around SEOs announcement and cut after SEOs announcement. 4. The firms who continue paying dividend for longer time before SEOs announcement would get less long term negative returns. The steadier dividend policy gets the better long term return. This finding is consistent with asymmetric information and signal theory that dividend does imply some information and the long term return would be affected. 4 Michaely, Richard, and Thaler (1995) proofed that dividend omissions would cause long term negative return. Loughran and Ritter (1995), Katherine and Affleck (1995), Brav, Geczy, and Paul (2000), Mitchell and Stafford (2000) proofed that issuing seasoned equity offering would long term negative return. 22

The result admits the signaling model presumption that dividend s convey information which can reduce the valuation uncertainty. Firms who execute steady dividend policy would have less asymmetric information. Consequently, this kind of firms would receive less negative returns by issuing SEOs. 23

Reference Asquith P. and D. W. Mullins, Jr., 1986, Equity issues and offering Dilution, Journal of financial Economics 15, 61-89. Barber, Brad M., and John D. Lyon, 1997, Detecting long-run abnormal stock returns: Theempirical power and specification of test statistics, Journal of Financial Economics43, 341-372. Bayless M. and Chaplinsky S., 1996, Is there a window of opportunity for seasoned equity issuance? Journal of Finance 51, 253-278 Brous P. A., 1992, Common stock offerings and earnings expectations: a test of the release of unfavorable information, Journal of Finance 40, 1517-1536. Bernard V. L. and J. K. Thomas, 1990, Evidence that stock prices do not fully reflect the implications of current earnings for future earnings, Journal of Accounting and Economics 13, 305-340. Chang Bin, 2006, Does the market react less negatively to dividend payers' seasoned equity offerings? Working paper. DeAngelo, H., L. DeAngelo,, and D. Skinner, 2003, Reversal of fortune, dividend signaling and the disappearance of sustained earnings growth, Journal of Financial and Quantitative Analysis 26, 181-199. Denis, D. J., D. K. Denis, and A. Sarin, 1994, The information content of dividend changes: cash flow signaling, overinvestment, and dividend clienteles, Journal of Financial and Quantitative Analysis 29, 567-587. D. Katherine S. and J. Affleck G., 1995, Underperformance in long-run stock returns following seasoned equity offerings, Journal of Financial Economics 38, 243-267. D Mello, R., O, Tawatnuntachai, and D. Yaman, 2003, Does the sequence of 24

seasoned equity offerings matter? Financial Management 2003, 59-86. Eckbo, B. E., Masulis, R. W., Norli O., 2000, Seasoned public offerings: resolution of the new issues puzzle Journal of Financial Economics 56, 251-291. 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., 1998, Market efficiency, long-term returns, and behavioral finance, Journal of Financial Economics 49, 283-306. Fama, Eugene F., and Kenneth R. French, 2001 Disappearing dividends: changing firm characteristics or lower propensity to pay? Journal of Financial Economics 60, 3-43. Frank H. Easterbrook, 1984, Two agency-cost explanations of dividends, The American Economic Review 74, 650-659. Healy, Paul M., and Krishna G. Palepu, 1990, Earning and risk changes surrounding primary stock offers, Journal of Accounting Research 28, 25-48. Jensen, Michael C, 1986, Agency costs of free cash flow, corporate finance, and takeovers, The American Economic Review 76, 323-329. Ritter J. R., 1991, The long-run performance of initial public offerings, Journal of Finance 46, 3-27. Korajczyk, R. A., D. J. Lucas, and R. L. McDonald, 1992, Equity issues with Time-varying asymmetric information, Journal of Financial and Quantitative Analysis 27, 397-417. Lintner J., 1956, Distribution of incomes of corporations among dividends, retained earnings, and taxes, The American Economic Review 46,97-113. Loderer, C. F. and D. C. Mauer, 1992, Corporate Dividends and Seasoned Equity Issues: An Empirical Investigation, Journal of Finance 47,73-112. Loughran, T., J. Ritter,, 1995, The new issues puzzle, Journal of Finance 50, 23-51. 25

Loughran, Tim, and Jay R. Ritter, 1997, The operating performance of firms conducting seasoned equity offerings, Journal of Finance 52, 1823-1850. Lyon, John D., Brad M. Barber, and Chih-Ling Tsai, 1999, Improved methodology for tests of long-run abnormal stock returns, Journal of Finance 54, 165-201. M. Bajaj, and A. M. Vijh., 1990 Dividend clienteles and the information content of dividend changes, Journal of Financial Economics 26, 193-219. Michaely, R., Thaler, R., Womack, K., 1995, Price reactions to dividend initiations and omissions, Journal of Finance 50, 573-608. Miller, Merton and Franco Modigliani, 1961, Dividend policy, growth, and the valuation of shares, Journal of Business 34, 411-433. Miller, Merton and Kevin Rock, 1985, Dividend policy under asymmetric information, Journal of Finance 40, 1031-1052. Mitchell, Mark L., and Erik Stafford, 2000, Managerial decisions and long-run stock price performance, Journal of Business 73, 287-329. 26

Table I Distribution of Seasoned Equity Offerings by Year and Industry The sample consists of 1333 SEOs events which occur during the period between 1987 and 2002 that are selected from Securities Data Company s New Issues Database. The share code in CRSP must be 10 or 11 means the sample excludes equity offerings by closed-end funds, real estate investment trusts (REITs), unit investment trusts, and American depositary receipt (ADRs). Panel A is the time distribution of the sample by event year. Panel B reports the SIC distribution of the sample by two-digit SIC code. Panel A. Time Distribution Fiscal Year End Number Percentage(%) 1987 17 1.28 1988 42 3.16 1989 37 2.78 1990 77 5.79 1991 72 5.42 1992 100 7.52 1993 117 8.8 1994 105 7.9 1995 156 11.74 1996 140 10.53 1997 119 8.95 1998 93 7 1999 128 9.63 2000 60 4.51 2001 61 4.59 2002 9 0.68 total 1333 100% 27

Panel B. SIC Distribution Industry Two-digit SIC codes Number Percentage (%) agriculture, forestry, and fishing 01-09, 20 17 1.28 mining 10-14 34 2.56 construction 15, 16, 17 12 0.9 manufacturing 29, 30, 31, 32,33, 34 57 4.29 wholesale trade 50, 51 70 5.27 retail trade 52-59 126 9.48 services 70-79 107 8.05 Health services 80 55 4.14 communication 48 54 4.06 Scientific Instruments 38 99 7.45 Transportation 37, 39, 40-42,44,45 66 4.97 Electric and electronic equipment 36 167 12.57 Paper and Paper Products 24, 25, 26, 27 27 2.03 Computer Hardware & software 35, 73 265 19.94 chemical products 28 146 10.99 the others 22, 23, 46, 47 14 1.53 total 1333 100 28

Table II Variable Characteristic of SEOs Firm The table shows the descriptive statistics of the SEOs firms. The sample consists of 1333 SEOs events which occur during the period between 1987 and 2002 that are selected from Securities Data Company s New Issues Database. Dividend-paying firms are defined as that pay cash dividend during the three quarters prior and current quarter on the seasonal equity offerings announcement, and the others are non-dividend payers. Panel A reports summary statistic of all sample events. Panel B is the summary statistic of dividend-paying firms, and Panel C is the summary statistic of non-dividend-paying firms. CAPX/TA is defined as capital expenditure divide into total asset. DLTT/TA is defined as long term debt divide into total asset. MV is defined as (total asset book value of equity + market value of equity). BMK is defined as (book value of equity / market value of equity). DTA is defined as (total debt / total asset). TA is defined as total asset. AGE is defined as the firm s operating years. Panel A. All SEOs Firm Mean Standard Dev. Q1 Median Q3 CAPX/TA(%) 8.2985 9.8053 2.7756 5.3309 9.7837 DLTT/TA(%) 17.1254 22.814 0.2487 7.2507 28.2704 ROA(%) -3.8261 26.5842-4.543 4.234 8.897 MV($ milloin) 1027.84 6363.5992 110.3149 232.6005 671.2586 BMK 0.3011 0.3916 0.1445 0.261 0.4126 DTA(%) 20.8894 24.1798 1.171 13.094 33.274 TA($ million) 382.076 1504.0632 36.404 82.045 236.906 logmv 2.4439 0.6043 2.0426 2.3666 2.8269 AGE 2.0539 2.3705 0 1 3 29

Panel B. Dividend-Paying Firms Mean Standard Dev. Q1 Median Q3 CAPX/TA(%) 8.167 9.8607 2.9518 5.4488 9.9266 DLTT/TA(%) 22.9829 18.8159 8.0387 21.641 34.2432 ROA(%) 7.1186 5.3915 3.812 6.4065 9.839 MV($ millions) 1903.198 6909.2936 239.2845 533.0445 1336.457 BMK 0.4518 0.3026 0.2689 0.4063 0.5871 DTA(%) 27.8475 19.4089 12.947 27.605 39.692 TA($ millions) 997.0705 2981.2505 124.355 318.4075 835.781 logmv 2.7714 0.5774 2.3789 2.7267 3.126 AGE 3.162 2.5561 1 3 5 Panel C. Non-Dividend-Paying Firms Mean Standard Dev. Q1 Median Q3 CAPX/TA(%) 8.3142 9.8028 2.7689 5.3201 9.7706 DLTT/TA(%) 16.4283 23.1523 0.1443 5.9221 25.8643 ROA(%) -5.1289 27.7759-7.436 3.889 8.703 MV($ millions) 923.6444 6290.5058 102.8938 208.5375 599.9948 BMK 0.2831 0.3972 0.1364 0.2476 0.3903 DTA(%) 20.0664 24.5599 0.916 11.38 31.331 TA($ millions) 308.8747 1195.8035 33.392 71.071 185.62 logmv 2.405 0.5958 2.0124 2.3192 2.7781 AGE 1.922 2.3135 0 1 3 30

Table III Five Year Buy-and-Hold Abnormal Returns for All SEOs Firms The table shows the buy-and-hold abnormal returns (BHARs) of all SEOs firms during from 1987 to 2002. The five-year abnormal returns are calculated by three different benchmarks; one is size and B/M ratio matching firm, one is CRSP equally-weighted Index, and the other is CRSP value-weighted Index. We use the Cross-sectional t-statistic and Skewness-adjusted t-statistic to test the significance of the mean value of BHARs. T-statistics are reported with ***, **, and * represent significance at the 1%, 5%, and 10% level each. 1 year 2 year 3 year 4 year 5 year A. based on size and B/M ratio matching firm Abnormal return (Mean) -0.28% -5.32% -6.17% -11.60% -11.57% Cross-sectional t-stat -0.185-2.298* -0.984-2.637** -2.383** Skewness-adjusted t-stat -0.186-2.049* -0.777-2.135* -2.357** B. based on CRSP equally-weighted Index Abnormal return (Mean) -2.99% -16.70% -23.37% -31.55% -30.06% Cross-sectional t-stat -1.377-5.778*** -6.078*** -7.013*** -4.741*** Skewness-adjusted t-stat -1.314-4.733*** -4.461*** -5.081*** -3.513*** C. based on CRSP value-weighted Index Abnormal return (Mean) -3.41% -16.52% -22.56% -26.02% -19.11% Cross-sectional t-stat -1.55-5.675*** -5.823*** -5.734*** -3.027** Skewness-adjusted t-stat -1.469-4.721*** -4.434*** -4.515*** -2.525* 31

Table IV Calendar-Time Fama and French Three-Factor Model Portfolio Regression for All SEOs Firms This table shows the portfolio abnormal returns for all SEOs firms which are based on the three-factor model of Fama and French (1993). The equation is as follow: Where dependent variable, Rpt, is the equally-weighted event portfolio return. Rft is the return on one-month Treasury bills in month t, is the return on a value-weighted market index in month t, SMB t R is the difference in the returns of a portfolio of the small and big stocks in months t, HML is the difference in the returns of a portfolio of high book-to-market stocks and low book-to-market stocks in the month t. pt Newey West t-statistics are reported in parentheses. ***, **, and * represent significance at the 1%, 5%, and 10% level each. R ( R R ) s SMB h HML ft i R i mt mt ft i i i i it t Holding Period Intercept (α) RMRF SMB HML R-squared 1 year -0.0016 1.253 1.057-0.6051 0.8371 (-0.58) (16.20)*** (10.33)*** (-6.20)*** 2 year -0.0041 1.3056 1.0003-0.3961 0.8321 (-1.54) (19.53)*** (10.07)*** (-4.60)*** 3 year -0.0019 1.3408 0.9623-0.2734 0.8233 (-0.76) (19.97)*** (10.46)*** (-3.45)*** 4 year -0.0023 1.3291 0.9725-0.1839 0.8276 (-0.99) (20.24)*** (11.71)*** (-2.46)** 5 year -0.0021 1.3204 0.9883-0.1413 0.8317 (-0.9) (20.26)*** (12.13)*** (-1.92)* 32

Table V Five Year Buy-and-Hold Abnormal Returns for Dividend-Paying Firms and non-dividend-paying Firms The table shows the buy-and-hold abnormal returns (BHAR) of dividend-paying firms and non-dividend-paying firms during from 1987 to 2002. Panel A presents the dividend-paying firms, and Panel B presents the non-dividend-paying firms. The five-year abnormal returns are calculated by three different benchmarks; one is size and B/M ratio matching firm, one is CRSP equally-weighted portfolio, and the other is CRSP value-weighted portfolio. We use the Cross-sectional t-statistic and Skewness-adjusted t-statistic to test the significance of the mean value of BHARs. T-statistics are reported with ***, **, and * represent significance at the 1%, 5%, and 10% level each. Panel A. Dividend-Paying Firms 1 year 2 year 3 year 4 year 5 year A. based on size and B/M ratio matching firm Abnormal return (Mean) -0.21% -2.59% 2.29% -4.12% 4.60% Cross-sectional t-stat -0.088-0.742 0.395-0.54 0.287 Skewness-adjusted t-stat -0.088-0.726 0.41-0.528 0.308 B. based on CRSP equally-weighted portfolio Abnormal return (Mean) 0.90% -5.35% -16.53% -28.96% -35.13% Cross-sectional t-stat 0.24-0.899-1.823* -3.029** -2.824** Skewness-adjusted t-stat 0.242-0.859-1.452-2.436** -2.167** C. based on CRSP value-weighted portfolio Abnormal return (Mean) 1.30% -2.58% -11.50% -21.53% -23.62% Cross-sectional t-stat 0.331-0.415-1.228-2.133** -1.814* Skewness-adjusted t-stat 0.334-0.408-1.077-1.883* -1.574* Panel B. Non-Dividend-Paying Firms 1 year 2 year 3 year 4 year 5 year A. based on size and B/M ratio matching firm Abnormal return (Mean) -0.19% -6.43% -7.56% -13.07% -12.99% Cross-sectional t-stat -0.121-2.736** -1.204-2.958** -2.669** Skewness-adjusted t-stat -0.121-2.395** -0.9-2.337** -2.637** B. based on CRSP equally-weighted Index Abnormal return (Mean) -3.45% -18.05% -24.19% -31.86% -29.46% Cross-sectional t-stat -1.446-5.723*** -5.806*** -6.494*** -4.244*** Skewness-adjusted t-stat -1.373-4.633*** -4.244*** -4.731*** -3.201** C. based on CRSP value-weighted Index Abnormal return (Mean) -3.97% -18.18% -23.88% -26.55% -18.57% Cross-sectional t-stat -1.643* -5.736*** -5.700*** -5.380*** -2.693** Skewness-adjusted t-stat -1.547* -4.693*** -4.287*** -4.232*** -2.271** 33