興櫃轉上市櫃公司股票抑價與財經新聞情緒分析之關聯性研究 - 政大學術集成

全文

(1)國立政治大學會計學系碩士論文. 興櫃轉上市櫃公司股票抑價與財經新聞情緒分析之關聯性研究政治大立. ‧ 國. 學. THE ASSOCIATION BETWEEN IPO. ‧. UNDERPRCING AND FINANCIAL NEWS. n. al. er. io. sit. y. Nat. – A SENTIMENT ANALYSIS APPROACH. i n U. Ch. v. engchi 指導教授：諶家蘭博士研究生：. 中. 華. 民. 國. 楊璧華. 一O四. 年. 撰. 六. 月.

(2) 謝辭. 首先，感謝大學摯友郁惠、以姍以及學長祥豪，因為那次的飯局和你們的幫忙，我才能進入自己夢想中的會研所學習。謝謝不在身邊但永遠聽我那麼多，給我那麼多支持的容溶、幸真和 Fish，謝謝有妳們一起分享生活。這兩年要感謝的人很多，謝謝我的指導老師－諶家蘭老師，從研究案一路跟隨老師學習，老師的認真一直是我心中的榜樣，更謝謝老師的指導，我才能完成論文。更要感謝筱芳學姐、巧意學姐和祐子的在資料蒐集上的大力幫忙，. 政治大. 論文才能順利完成。謝謝研究案的大家，和你們一起共事的時光，很開心又收. 立. 穫滿滿。. ‧ 國. 學. 兩年碩班生活匆匆但卻美好，謝謝冠穎、怡慧和駿廷，與你們一起工作的這些日子以來，學到好多，冠穎的從容、怡慧的開朗與認真，駿廷的用心。謝. ‧. 謝熊曦和柏廷給我論文上的建議以及各種精神上的支持！你們鼓勵的話，我會. sit. y. Nat. 永遠記得。謝謝九州小畢旅的大家，那一次的旅行很棒，仍讓我回味不已，希. al. er. io. 望未來也能常常約出來關心彼此！謝謝熊曦、怡慧和昀耘，在你們面前，我可. v. n. 以是最真實的小璧，謝謝這兩年你們的陪伴與包容。謝謝每一位會研的老師及. Ch. engchi. i n U. 同學，這兩年有幸與大家一起學習成長，無盡的感恩，點滴在心頭。感謝我的爸爸、媽媽，我的家人，從小到大對我的栽培與教育，無條件地給予任何支持，沒有你們沒有我，一路上有你們，我才能專心求學，享受知識的美好。. 楊璧華謹至於國立政治大學會計學研究所 2015 年 6 月.

(3) 摘要. 本研究以民國 99 年至 103 年間於台灣證券交易所以及櫃買中心掛牌之新上市公司為樣本，並扣除由上櫃轉上市公司的部份，共計 241 間公司。由於投資人在公司上市前較難以取得公司相關資訊、或是取得資訊成本極高，因此大部分資訊來自新聞媒體的報導。因此本研究目的在於媒體報導的正負面情緒與興櫃轉上市櫃公司的期初報酬率的關聯性，並以實證進行研究，以了解媒體報導密集度及情緒程度影響期初報酬率(初次公開發行折價幅度)之關係。. 政治大向媒體情緒程度分數。本研究實證結果顯示，媒體報導帶給投資人之情緒越正立. 本文以客觀量化之方式計算新聞內正向字詞與負向字詞，並據以計算為正. ‧ 國. 學. 向，期初報酬率越高。在上市日前一個月的正向新聞報導數量，對期初報酬率也有正面的影響；並將樣本區分為電子業及非電子業，電子業在上市日前一個. ‧. 月的正(負)向新聞報導數量，對期初報酬率有正(反)面的顯著影響；非電子業則. y. sit. n. al. er. io. 之影響。. Nat. 與正向媒體情緒、正(負)向新聞報導數量數較無顯著相關，反而受大盤報酬率. Ch. engchi. 關鍵字：折價幅度、媒體報導、情緒分析. i n U. v.

(4) Abstract This study focuses on the relationship between the sentiment of the media coverage and IPO underpricing based on the IPOs firm listing on TWSE corporations or GTSM corporations. This study investigates whether the media were responsible for the phenomenal rise and fall in the market value of IPO shares from 2010 to 2014. We do not focus on what drives the media coverage of IPO firms, but we focus on the effect of pre-IPO. 政治大 Recently, sentiment approach have been a useful tool for extract the text 立. media coverage on the first day's return.. information to the numeral data and would be easy to compare and calculate. To avoid. ‧ 國. 學. the subjective judgment bias, we use sentiment dictionary-based approach to convert. ‧. the qualitative news to quantitative measures. This study uses the positive and negative. sit. y. Nat. words to calculate positive sentiment score of the media, and use the positive and. io. er. negative words to classify the news items into positive or negative news. In summary, this positive sentiment of media is significantly related to initial. al. n. v i n C hstudy also indicatesUa significantly result that initial return (IPO underpricing), and this engchi return is highly related to market return than sentiment of media in non-electronics industry. However, as our expected, the number of positive (negative) news reports has a positive (negative) influence on the initial return, in the electronics industry.. Keywords：IPO underpricing，Media Coverage，Sentiment Analysis.

(5) Table of Contents INTRODUCTION ................................................................................................. 1 1.1 RESEARCH PURPOSE AND MOTIVATION ............................................... 1 1.2 RESEARCH ISSUE......................................................................................... 4 1.3 RESEARCH PROCESS .................................................................................. 5 2. LITERATURE REVIEW ........................................................................................... 7 2.1 IPO UNDERPRICING AND ASYMMETRIC INFORMATION ................... 7 2.2 IPO UNDERPRICING AND SIGNALING THEORIES ................................ 9 2.3 BOOK-BUILDING METHOD AND INFORMATION PRODUCTION 1.. THEORY.............................................................................................................. 10 2.4 IPO UNDERPRICING AND OPTIMISTIC INVESTORS .......................... 12 2.5 LITERATURE REVIEW SUMMARY FOR THE THEORIES OF IPO UNDERPRICING ................................................................................................ 15 2.6 EMPIRICAL RESEARCHES ABOUT MEDIA COVERAGE AND IPO UNDERPRICING ................................................................................................ 20. 立. 政治大. ‧ 國. 學. ‧. 2.7 SENTIMENT ANALYSIS ............................................................................. 25 3. SENTIMENT ANALYSIS MODEL........................................................................ 27 3.1HYPOTHESES DEVELOPMENT ................................................................ 27 3.1.1 THE RELATIONSHIP BETWEEN IPO UNDERPRICING AND THE. n. al. er. io. sit. y. Nat. SENTIMENT OF MEDIA COVERAGE .................................................... 27 3.1.2 THE RELATIONSHIP BETWEEN IPO UNDERPRICING AND THE NUMBER OF POSITIVE (NEGATIVE) NEWS REPORTS ..................... 28 3.2 DATA COLLECTION ................................................................................... 30 3.2.1 SAMPLE SELECTION .............................................................................. 30 3.2.2 SAMPLE DESCRIPTION .......................................................................... 31 3.3 REASERCH METHOD................................................................................. 35 3.3.1 APPROACHES TO MEASURE THE SENTIMENT POLARITY IN NEWS ITEM ....................................................................................................... 35 3.3.2 LYDIA SENTIMENT ANALYSIS ............................................................. 37. Ch. engchi. i n U. v. 3.4 REGRESSION MODEL................................................................................ 38 3.4.1 DEPENDENT VARIABLE ........................................................................ 41 3.4.2 INDEPENDENT VARIABLES .................................................................. 42 3.4.3 CONTROL VARIABLES ........................................................................... 43 4. RESEARCH RESULTS AND ANALYSIS ............................................................. 44 4.1 DESCRIPTIVE STATISTICS ....................................................................... 44 4.2 EMPIRICAL RESULTS ................................................................................ 53 4.2.1 REGRESSION RESULTS FOR H1-MODEL (1) TO (4) .......................... 53.

(6) 4.2.2 REGRESSION RESULTS FOR H2-MODEL (5) ...................................... 58 5. CONCLUSION AND DISCUSSION .................................................................. 61 5.1 RESEARCH CONTRIBUTION AND CONCLUSION ............................... 61 5.2 LIMITATION AND FUTURE RESEARCH WORK .................................... 62 REFERENCES ............................................................................................................ 64. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. i n U. v.

(7) List of Tables TABLE 2-1 LITERATURE REVIEW SUMMARY FOR THE THEORIES OF IPO UNDERPRICING ................................................................................................ 15 TABLE 2-2 EMPIRICAL RESEARCHES ABOUT MEDIA COVERAGE............... 20 TABLE 3-1 SAMPLE DISTRIBUTION ..................................................................... 33 TABLE 3-2 NEWS DISTRIBUTION- BY INDUSTRY AND NEWS CLASSIFICATION.............................................................................................. 34 TABLE 3-3 FRAMEWORK OF THE MODEL ......................................................... 38 TABLE 4-1 DESCRIPTIVE STATISTICS (THE PRE-IPO PERIOD) PANEL A. FULL SAMPLE ................................................................................................... 47 TABLE 4-1 DESCRIPTIVE STATISTICS (THE PRE-IPO PERIOD) PANEL B. SAMPLE DIVIDED BY INDUSTRY ................................................................. 48 TABLE 4-2 DESCRIPTIVE STATISTICS (ONE MONTH PRIOR TO THE ISSUE DAY) PANEL A. FULL SAMPLE ...................................................................... 49. 立. 政治大. ‧ 國. 學. ‧. TABLE 4-2 DESCRIPTIVE STATISTICS (ONE MONTH PRIOR TO THE ISSUE DAY) PANEL B. SAMPLE DIVIDED BY INDUSTRY .................................... 50 TABLE 4-3 PEARSON CORRELATION (THE PRE-IPO PERIOD) ........................ 51 TABLE 4-4 PEARSON CORRELATION (ONE MONTH PRIOR TO THE ISSUE. n. al. er. io. sit. y. Nat. DAY) .................................................................................................................... 52 TABLE 4-5 REGRESSION RESULT FOR H1-MODEL (1) TO (4) .......................... 55 TABLE 4-6 REGRESSION RESULT FOR H1-MODEL (1) AND (2) AT INDUSTRY LEVEL ................................................................................................................. 57 TABLE 4-7 REGRESSION RESULT FOR H2-MODEL (5)...................................... 59 TABLE 4-8 REGRESSION RESULT FOR H2-MODEL (5) AT INDUSTRY LEVEL .............................................................................................................................. 60. Ch. engchi. i n U. v.

(8) List of Figures FIGURE 1-1 RESEARCH PROCESS .......................................................................... 5 FIGURE 3-1 HYPOTHESES STUCTURE................................................................. 29 FIGURE 3-2 THE APPROACH OF NEWS SENTIMENT CLASSIFICATION ....... 37. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. i n U. v.

(9) List of Appendixes APPENDIX 1. DEFINITION OF VARIABLES ......................................................... 68 APPENDIX 2. POSITIVE WORD LIST .................................................................... 70 APPENDIX 3. NEGATIVE WORD LIST .................................................................. 72. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. i n U. v.

(10) 1. INTRODUCTION 1.1 RESEARCH PURPOSE AND MOTIVATION Corporations may raise capital in the primary market by way of an initial public offer, rights issue or private placement. An Initial Public Offerings (IPOs) is the selling of securities to the public in the primary market. The main initial public offerings (IPOs) method are book-building, auction method, and public offer. Most companies in Taiwan securities market are using the combination way of book-building and public offer. The book-building method also has the advantage of reducing the degree of underpricing.. 政治大. In general, the book-building method is essentially a process used by companies raising. 立. capital through Public Offerings, both Initial Public Offers (IPOs) or Follow-on Public. ‧ 國. 學. Offers (FPOs) to bid price and demand discovery. It is a mechanism where, during the period for which the book for the offer is open, the bids are collected from investors at. ‧. various prices, which are within the price range specified by the issuer and underwriter.. y. Nat. io. sit. The process is directed towards both the institutional as well as the retail investors. The. n. al. er. offer price is determined after the bid closure based on the demand generated in the. i n U. v. process. Book-building models beginning with Benvensite and Spindt (1989) and. Ch. engchi. Benveniste and Wilhelm (1990) argue that the underwriter’s control of both price and allocations may be used to induce investors to reveal their private information. These explanations focus on asymmetric information and the difficulties with establishing a right price value for new shares. However, the book-building bids and allocation data are not publicly accessible. It is difficult for us to observe how the final shares are allocated, although we know the underwriter have the preference for specific informed investors. Hence, it is difficult to research on the implications of book-building model. Accordingly, we try to use the information flows as the key factor to observe this pricing process. 1.

(11) In addition, emerging markets are usually operated in an environment of information opaqueness as pointed out by Morck, Yeung, and Yu (2000), stock prices in these markets are quite sensitive to political events and rumors. Furthermore, domestic individual investors are the main group in the Taiwan market in that they dominate majority of total security trading, but they rely on noise in the market. Consistent with recent studies, IPO firms tend to show themselves as those with good prospects. These issuers also intend to make use of every opportunity to maximize their own personal wealth. Arbel and Strebel (1983) suggest that individual investors. 政治大 Especially, unlike the regulatory environment of equity offering in the United States, 立 are usually uninformed, poorly trained, and inclined to rely on noise in the market.. there is no restriction on news announcements of issuers around IPOs in Taiwan. It also. ‧ 國. 學. allows IPO firms to release news that can generate favorable views at their discretion.. ‧. For instance, a firm can announce its potential operation and business strategies via. sit. y. Nat. news. These news shows that the firms have an attempt to present their positive image. io. er. and attract investors. Therefore, examining financial news of IPO firms in Taiwan allows us to investigate whether the investors are influenced the sentiment of the media.. al. n. v i n C habout the firm’s Ufundamentals information engchi. News carries. and qualitative. information influencing expectation of market participants. The news media de facto influences stock prices to some extent. There is also evidence that investors are not only subject to the sentiment of related news but also public opinions. Comparing to forecasted earnings disclosed by IPO firms, news regarding these firms announced through media are more qualitative in nature. In this thesis, we focus on the media attention before the IPO day. By going through the book-building process, issuers are attempting to attract the investors’ attention. Also, underpricing may be a way to induce the investors to know this particular company, and pay attention on the initial public offerings. We focus on 2.

(12) examining the relationship between media coverage and underpricing. In the past researches, we find that the relationship between media coverage and price revision is consistent with the information production theory. Liu, Sherman and Zhang (2009) indicate that book-building model predict that underpricing will be concentrated in high demand offering, in order to make it easier to induce truthful revelation of investors’ private information even when that information will be used to raise the offer price for high demand offerings. Before discussing the relation between the media coverage and underpricing, they define investors’ demand is quiet relatively to the offer price. With. 政治大 is revised upwards from the midpoint of the initial filing range. Whereas, when demand 立. the revelation of private information, when the investors’ demand is high, the offer price. is low, the offer price is revised downwards. Price revision will be used as an important. ‧ 國. 學. control variable mentioned in section 3.4.. ‧. This thesis is closely related to Liu et.al (2009), which established that more pre-. y. Nat. IPO media coverage is strongly related to the liquidity. Liu et al. (2009) propose that. er. io. sit. the media coverage as a proxy for the investors’ attention. They count the number of articles from these media sources covering the IPO companies during the window.. al. n. v i n C h across firm, theyUstandardize the media coverage Since the length of the window varies engchi. measure into a per month basis and use it in their empirical analyses. The big difference between this thesis and the research of Liu et.al (2009) is that Liu et.al (2009) do not attempt to categorize the news announcement as “positive” or “negative”. This study investigates whether the sentiment of media is responsible for the phenomenal rise and fall in the market value. Furthermore, we look at both the number of the news announcements and their sentiment type. This study motivated by the recent literature that investigates the role of investor sentiment (Neal and Wheatley 1998 and Baker and Wurgler 2003). We focus on whether the IPO firms can engage in other qualitative types of reporting to influence 3.

(13) investors’ perception, and their demand for the new issue shares. In order to test this question, this study collects the news items from the largest internet news database and categorize them (positive, negative, or neutral). Although the textual information is an important component of the news. In our sample, the news covers a wide of topic, such as the company’s recent financial performance, business strategies, strategic alliances, competitive position within the industry, and so on, this study does not focus on what type of the news have the relation with underpricing. On the other hand, there is a key characteristic of the news is the effect of the sentiment polarity in news that we focus on.. 政治大 To avoid the subjective judgment bias, we use sentiment dictionary-based 立. approach to convert the qualitative news into quantitative measures. Azar (2009) argue. ‧ 國. 學. that through the sentiment analysis process, information is converted from a textual. sit. y. Nat. allows information to be easily summarized and compared.. ‧. form to a numeric form. To convert the text information is a useful way because it. io. er. In summary, using this sentiment approach to measure whether the media coverage is related to underpricing or not help us to know the positive or negative information. al. n. v i n Ctohknow the investorUsentiment for the IPO firms. affect the offer price, also help us engchi 1.2 RESEARCH ISSUE. According to the research purpose and motivation, the research questions of this study is as follows: 1. Does the sentiment of media coverage is associated with underpricing?. 4.

(14) 1.3 RESEARCH PROCESS The research process in this study is shown as following:. FIGURE 1-1 RESEARCH PROCESS. Research Background and Research Problems Identification. 政治大. 立 Literature Review. ‧. ‧ 國. 學. Nat. n. al. er. io. sit. y. Research Method. Ch. engchi. i n U. Empirical Results and Analysis. Discussion and Conclusion. 5. v.

(15) In Figure 1-1, this study constructs the research process. First, after deciding the research topic and defining the research background, this study identifies main questions about this research. Second, this study searches relevant literatures and theories and performs research hypotheses. This study collects and understands raw data and information of news items, and build up the research method. Eventually, this study uses Ordinary Least Squares (OLS) regression to test the relation between the sentiment of media and IPO underpricing. The empirical results of our models are discussed and form the conclusion.. 政治大 literatures and theories. Section 3 introduces the research method, the data set and all 立. The remainder of this study is organized as follows. Section 2 introduces relevant. the variables used in the model. In section 4, it presents the empirical results and the. ‧ 國. 學. relation between the media coverage and underpricing. Finally, section 5 concludes.. ‧. n. er. io. sit. y. Nat. al. Ch. engchi. 6. i n U. v.

(16) 2. LITERATURE REVIEW 2.1 IPO UNDERPRICING AND ASYMMETRIC INFORMATION Most models of IPO underpricing are based on asymmetric information. There are two explanations of underpricing in the IPO literature. Baron (1982) uses a model of underpricing where issuers delegate the pricing decision to underwriters. Baron (1982) presents that the investment banker is better informed about the capital market than is the issuer, and the issuer cannot observe the distribution effort expended by the banker. If the banker is better informed than the issuer, however, the issuer will not only. 政治大 contract ,but also the advising 立 services under a delegation contract. Under such a. have a demand for the banker's distribution services under the pure distribution. ‧ 國. 學. contract, the offer price decision is delegated to the banker who sets the offer price based on his superior information about the capital market. The issuer must compensate. ‧. the banker for the use of his information, so the banker shares in the gains from his. sit. y. Nat. superior information. In the example, the optimal offer price is below the first-best offer. n. al. er. io. price indicating that new issues would be underpriced when the banker is better. i n U. v. informed than the issuer. Furthermore, Investment bankers find it less costly to market an IPO that is underpriced.. Ch. engchi. A best known study of asymmetric information about the IPO underpricing is modeled by Rock (1986). Rock presents the privileged investors whose information is superior to the issuer as well as all other investors. In the model of Rock (1986) is that the issuer himself is also uninformed. Therefore the IPO offer price set by him cannot depend on firm type. This means that the offer price is not informative to any investor in the IPO, so that uninformed investors face severe adverse selection in the Rock (1986). Rock theorizes “winner’s curse”, “informed investors” like investment institution have superior information about true value to the stock after the listing；on 7.

(17) the contrary, “uninformed investors” have more probability of buying the inferior offering. This information asymmetry may lead to a “lemons problem”, uninformed investors must lose money in the long run and will want to withdraw from the stock market. Keeping them in the market, therefore, requires an additional premium-the average underpricing of all IPOs. Consequently, issuers must underprice the securities to attract and compensate uninformed investors for the probability of overestimated offer price. Beatty and Ritter (1986) extend Rock’s theory and find evidence that risk does. 政治大 first being the number of uses of proceeds listed in the new issue prospectus. The SEC 立. affect after-market returns. They use two different proxies for ex-ante uncertainty, the. requires issuers that they deem to be more speculative to provide more detailed. ‧ 國. 學. explanations of their expected uses for the proceeds from the offering. Hence, those. ‧. issues with more detailed listings in their filings may be d more speculative. The second. sit. y. Nat. risk proxy used by Beatty and Ritter is the inverse of the gross proceeds for the given. io. er. issue. Their justification for this proxy is the empirical regularity that smaller issues are generally more speculative than larger issues. Beatty and Ritter find that both the. al. n. v i n number of uses for the proceedsCand the inverse of theUgross proceeds are statistically hengchi significant explanatory variables of initial returns. However, the r-squared for their multivariate regression was low. Consequently, if an uninformed investor is allocated shares in an initial public offering, there is a greater than usual chance that the issue will start trading at a discount in the aftermarket. In other words, for an uninformed investor, the expected return conditional upon being allocated shares is less than the expected return conditional upon submitting a purchase order. But an uninformed investor will participate in the market only if the expected return conditional upon being allocated shares is nonnegative. This can only happen if, on average, issuers underprice their shares. The 8.

(18) owners of a firm going public, who typically have a large proportion of their wealth invested in the firm, would be willing to pay this price if they are sufficiently riskaverse. 2.2 IPO UNDERPRICING AND SIGNALING THEORIES Welch’s (1989) model, as well as Grinblatt and Hwang’s (1989) and Allen and Faulhaber (1989), formalize the notion that good firms underprice to “leave a good taste in investors’ mouths.” The hypothesis is that the owner’s incentive to leave a good taste is the possibility of coming back to the market for the sale of additional securities on. 政治大 subsequent issuance of equity or debt in the secondary market. The entrepreneur of a 立 more favorable terms. Welch’s model explicitly accounts for the possibility of. high-quality firm will underprice to distinguish his firm from the low-quality firm. He. ‧ 國. 學. will be rewarded at the time of the seasoned issue by a higher price for the shares. The. ‧. underpricing is a credible signal if the imitation costs for the low-quality firm are high. sit. y. Nat. enough. In fact, the entrepreneur faces a joint decision of how much to underprice and. io. er. whether to reenter the market at a later time for the subsequent issue of equity or debt. In addition, Welch’s model implies that good firms, which underprice more,. al. n. v i n C hresponse at the timeUof the seasoned issue. experience a less unfavorable price engchi. Allen and Faulhaber (1989) classify firms fall into two types: good firms and bad. firms. A firm's type can change through time, depending on its performance. The only basis investors have for their belief about the firm's prospects is the price of its IPO. Their analysis distinguishes between a separating equilibrium, where good firms signal their type to investors with a low IPO, and a pooling equilibrium, where no underpricing occurs, and hence investors cannot tell good and bad firms apart. Pooling is more profitable for good firms if they are likely to remain good, and signaling is more profitable if they are likely to get worse. Empirical evidence confirms that underpricing can signal favorable prospects for the firm, and that it is temporary, industry-specific, 9.

(19) and associated with improvements in the profitability of entry. 2.3 BOOK-BUILDING METHOD AND INFORMATION PRODUCTION THEORY Benveniste and Spindt (1989) develop a theory of underwriting that underwriters as the role to improve the economic efficiency of the initial public offerings market. They express that the underwriters underprice to compensate investors with positive information about the value of the stock for truthful disclosure of their private information. Based on the theory, the offer price may be partial adjusted to meet the. 政治大 information. This means, offer price may be fully adjusted to the market value under 立. market value, the unadjusted part is to compensate the investors with private. the public information.. ‧ 國. 學. Investment bankers tend to favor their large and established customers in. ‧. allocating shares of initial public offerings. These customers are also more likely to be. sit. y. Nat. the better informed investors. The retail (or uninformed) investors will face an. io. er. allocation bias whether the institutional investors receive larger allocations in the better IPOs because they bid more for them (as a result of their superior information) or. al. n. v i n because they are favored by the C investment bankers. Prior h e n g c h i U knowledge of the absence of the informed investors reduces the winner’s curse problem and consequently the need for underpricing.. From the research of Sherman and Titman (2002), we know the book-building process could be described as consisting of three steps. The investment bank first decides which investors will be invited to evaluate and perhaps buy the issue. Next, investors evaluate the issue and provide the investment bank with preliminary indications of their demand for the issue. Finally, the investment bank prices the issue and allocates shares to investors, generally allocating more shares to investors who indicate higher levels of demand. The book-building method is often criticized because, 10.

(20) in contrast to the other methods, underwriters generally exclude investors from the bidding process, provides the new shares to reward their favorite clients. This is especially controversial. Sherman and Titman’s (2002) model of the book-building process, an extension of earlier models, provides an explanation for why investment banks limit the number of investors who are invited to participate in an IPO. The model is based on two premises. The first is that collecting information is costly, which is a fairly standard and intuitive assumption. The second is that there are economic benefits associated with. 政治大 secondary market. This implies that, ex ante, the entrepreneur would like a mechanism 立 generating information in order that the IPOs are priced more accurately in the. for going public that prices the firm as accurately as possible. As Benveniste and Spindt. ‧ 國. 學. (1989) and others have emphasized, the investment bank’s pricing policy affects the. ‧. incentives of investors to truthfully reveal their private information. In particular, if. sit. y. Nat. investment bankers do not sufficiently underprice those new issues with the most. io. er. favorable information, investors will tend to underreport their information. When information is costly for investors to obtain, there is a second rationale for. al. n. v i n Ch underpricing new issues. Investors face a moral hazardU problem because they know that engchi. the price of the issue will be based on all reported information. Rather than purchasing his or her own signal, each investor may be tempted to free-ride on the information of others. IPOs must be underpriced sufficiently to offset this problem and to induce all investors to collect information. Sherman and Titman (2002) explore the book-building process, which is increasingly being used throughout the world to distribute IPOs. By endogenizing the information acquisition process, they can examine how the underwriter selects the investors that participate in the offering and how the price and allocation schemes are determined. They demonstrate the tradeoff that the underwriter faces when increasing 11.

(21) the size of the investor pool. If more investors are invited to participate, the issue can be expected to price more accurately, but increased participation increases underpricing. When information is costless, the optimal number of participating investors is infinite and underpricing approaches zero. When information is costly, the level of underpricing is determined by the desire for information. For firms with the most to gain from accurate pricing, more investors will be invited to participate in the offering and underpricing will be greater. 2.4 IPO UNDERPRICING AND OPTIMISTIC INVESTORS. 政治大 IPOs. Cook, Kieschnick and Van Ness (2006) follow up the research model from 立 Another strand of the literature focuses on the impact of optimistic investors on. Derrien (2005) and Ljungqvist, Nanda, and Singh (2006). Derrine (2005) and. ‧ 國. 學. Ljungqvist et al. construct models in which optimistic investors affect IPO prices and. ‧. positive IPO returns. Their models imply that sentiment investor (noise trader) demand. y. Nat. is the primary determinant of price run-ups in initial IPO trading. Sentiment investors. er. io. sit. are willing to overpay for the firm’s shares at the time of IPO. Consequently, one should expect retail investor trading to be positively correlated with the pre-offer publicity. al. n. v i n measure if publicity draws theseC investors into the IPOUmarket. hengchi. While the models of Derrien (2005) and Ljungqvist et al. (2006) suggest that the. issuer and regular customers of an investment banker benefit from the presence of sentiment investors, neither model considers that an investment banker might promote an issue in such a way as to induce sentiment investors into the market for an IPO. Such a possibility, however, is consistent with the premise that investment bankers act as marketing agents for issuers. There is also another research supporting this premise, Degeorge et al. (2004) find in the French IPO market that book-built issues attract more press than auctions, but only after the book-building route is selected. They also find that analysts affiliated with the lead underwriter issue more frequent and favorable 12.

(22) recommendations than they would under an auction process. These favorable recommendations occur even following poor stock performance. While Cook et. al (2006) focus on the role of publicity, they are aware that the methods of marketing IPOs to retail investors are changing as a result of the development of the Internet. For example, road show information is increasingly being made available to retail investors via the website. Cook et al. (2006) argue that the process of promoting new security issues is an important feature of security issuance. In fact, it is prior investment banker promotional. 政治大 purpose of these laws is not to prohibit the promotion of securities, but “to assure the 立. behavior that triggered the Blue Sky laws and the Securities Act of 1933. The basic. availability of adequate reliable information about securities which are offered to the. ‧ 國. 學. public” (Ratner 1998). ‧. Barber and Odean (2002) reveal one method that might be used by investment. sit. y. Nat. bankers to promote an issue to attract retail investors and enhance their bullishness on. io. er. an IPO. Using brokerage records, Barber and Odean find that retail investors are more likely to purchase attention-grabbing stocks. Similarly, Tetlock (2007) analyzes the. al. n. v i n affect of the Wall Street JournalC (2003) “Abreast of theUMarket” column and finds that hengchi either the media report investor sentiment before the sentiment is fully incorporated. into market prices or the media directly influence investors’ attitudes toward securities. In addition, Frieder and Subrahmanyam (2005) determine that individual investors are more likely to hold stock in highly visible companies. These results suggest that an investment banker's efforts to promote an IPO through increased media coverage increase retail interest in that stock. In summary, about the research result of Cook et. al (2006) as following: Consistent with their first prediction, they find a positive and significant correlation between retail trading activity during the first day of trading in an IPO and 13.

(23) the IPO’s pre-issue publicity. Consistent with their second prediction, pre-issue publicity is positively correlated with upward revisions in IPO offer prices and offer price valuations that are above comparable firms in their industry. In addition, they find that insider wealth gains exceed their dilution losses when more pre-issue publicity is associated with their IPO. Consistent with their third prediction, initial IPO returns are positively correlated with pre-issue publicity. And finally, consistent with their fourth prediction, they find that investment banker compensation is positively and significantly correlated with pre-issue publicity. Reinforcing the importance of. 政治大 switch lead investment bankers when they are effective at promoting their IPO but, 立 marketing to issuers and investment bankers, they find that issuers are less likely to. when they do switch, they often are able to increase the pre-offer publicity associated. ‧ 國. 學. with their seasoned equity offering (SEO).. ‧. n. er. io. sit. y. Nat. al. Ch. engchi. 14. i n U. v.

(24) 2.5 LITERATURE REVIEW SUMMARY FOR THE THEORIES OF IPO UNDERPRICING The following table 2-1 summarize the literature review for the theories of IPO underpricing: TABLE 2-1 LITERATURE REVIEW SUMMARY FOR THE THEORIES OF IPO UNDERPRICING. 治 Result 政 Research 大. IPO UNDERPRICING AND ASYMMETRIC INFORMATION Author. Research Methodology. Baron (1982). The asymmetric information arises from that the In the example, the optimal offer price is below the first-best offer price. 立. ‧ 國. informed than the issuer. Furthermore, Investment bankers find it less costly to market an IPO that is underpriced.. ‧. ROCK (1986). indicating that new issues would be underpriced when the banker is better. 學. banker is better informed than the issuer.. The asymmetric information arises from that the Rock theorizes "winner's curse", issuers must underprice the securities to. sit. y. Nat. privileged investors whose information is compensate uniformed investor.. io. investors.. n. al. er. superior to the issuer as well as all other. Beatty (1986). and. i n U. v. Ritter They extend Rock’s theory, and introduce ex- They prove there is a positive relation between the ex-ante uncertainty about. Ch. engchi. ante uncertainty to measure the degree of initial public offering's value and expected initial return. asymmetric. information.. Their justification for this proxy is the empirical regularity that smaller issues. They regard the number of use of proceed listed are generally more speculative than larger issues. Beatty and Ritter find that in prospectus as a proxy for ex ante uncertainty. both the number of uses for the proceeds and the inverse of the gross proceeds are statistically significant explanatory variables of initial returns. 15.

(25) IPO UNDERPRICING AND SIGNALING THEORIES Author. Research Methodology. Research Result. Welch’s (1989). The hypothesis is that the owner’s incentive to “leave a good taste in investors’ mouths” leave a good taste is the possibility of coming Welch’s model implies that good firms, which underprice more, experience. 政治大. back to the market for the sale of additional a less unfavorable price response at the time of the seasoned issue. securities on more favorable terms.. 立. Allen and Faulhaber The new firm itself possesses the best The analysis distinguishes between a separating equilibrium, where good. ‧ 國. 學. information about its own prospects. Good firms firms signal their type to investors with a low IPO, and a pooling equilibrium, use a low IPO price thus signaling to investors where no underpricing occurs, and hence investors cannot tell good and bad. ‧. that they expect to recoup this loss in their later firms apart. Pooling is more profitable for good firms if they are likely to performance, while bad firms cannot afford such remain good, and signaling is more profitable if they are likely to get worse.. y. Nat. io. well.. sit. signaling, since they do not expect to perform Empirical evidence confirms that underpricing can signal favorable prospects for the firm, and that it is temporary, industry-specific, and associated with improvements in the profitability of entry.. n. al. er. (1989). Ch. engchi. 16. i n U. v.

(26) BOOK-BUILDING METHOD AND INFORMATION PRODUCTION THEORY Author Benveniste Spindt (1989). Research Methodology. Research Result. and They model the premarket as an auction, conducted by Based on the theory, the offer price may be partial adjusted to meet the underwriter, in which investors understand how the market value, the unadjusted part is to compensate the investors. 政治大 Therefore, offer price may be fully adjusted to the market value under. their indications of interest affect the offer price and the with private information. stock allotments they receive.. 立. IPO offer prices must be set low to provide profit to the public information.. ‧ 國. 學. compensate investors for revealing information. The The analysis yields a number of empirical implications, including that amount of compensation required depends on how new issues will be underpriced and that distributional priority will be. ‧. much investors may expect to profit by hiding the given to an underwriter's regular investors information. Clearly, this depends directly on the extent. y. Nat. al. er. io. lower expected offer price.. sit. to which withholding positive information results in a. n. Sherman and Titman The model is based on two premises. The first is that Increased participation increases underpricing. (2002). Ch. i n U. v. collecting information is costly, which is a fairly When information is costly, the level of underpricing is determined. engchi. standard and intuitive assumption. The second is that by the desire for information. For firms with the most to gain from there are economic benefits associated with generating accurate pricing, more investors will be invited to participate in the information in order that the IPOs are priced more offering and underpricing will be greater. accurately in the secondary market.. 17.

(27) IPO UNDERPRICING AND OPTIMISTIC INVESTORS Author. Research Methodology. Research Result. Derrien (2005). Sentiment investor (noise trader) demand is the In Derrien (2005) model, the information consists two types of signals: first, primary determinant of price run-ups in initial private signals about the firm’s fundamental value that the underwriter has. 政治大 overpay for the firm’s shares at the time of IPO (1989) model. Second, a public signal about the sentiment investors by 立 offering the firm’s shares. The underwriter, who is in charge of choosing the IPO trading. Sentiment investors are willing to to extract from informed investors followed the Benvesiste and Spindt’s. ‧ 國. 學. the shares trade at a lower price than the fundamental value per share of the. ‧. firm obtained by aggregating private signals from informed investors, but lower than the price sentiment investors are willing to pay.. y. Nat. Kieschnick Investment bankers have an incentive to Consistent with their first prediction, they find a positive and significant. sit. Cook,. IPO offer price, has to provide aftermarket price support, which is costly if. al. IPO and the IPO’s pre-issue publicity. Consistent with their second. n. into the market for it.. er. io. and Van Ness (2006) promote an IPO to induce sentiment investors correlation between retail trading activity during the first day of trading in an. Ch. i n U. v. They expect that the promotional efforts of prediction, pre-issue publicity is positively correlated with upward revisions investment. bankers. should. influence. i etheningIPO c hoffer prices and offer price valuations that are above comparable firms. compensation of investment bankers, the in their industry. In addition, they find that insider wealth gains exceed their valuation of an IPO, its initial returns and dilution losses when more pre-issue publicity is associated with their IPO. trading, the wealth gains of insider shareholders, Consistent with their third prediction, initial IPO returns are positively. 18.

(28) and the likelihood that an issuer switches correlated with pre-issue publicity. And finally, consistent with their fourth investment bankers for a subsequent seasoned prediction, they find that investment banker compensation is positively and equity offering.. significantly correlated with pre-issue publicity. Reinforcing the importance of marketing to issuers and investment bankers, they find that issuers are less. 立. 治to switch lead investment bankers when they are effective at promoting 政 likely their IPO but,大 when they do switch, they often are able to increase the preoffer publicity associated with their SEO.. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. 19. i n U. v.

(29) 2.6 EMPIRICAL RESEARCHES ABOUT MEDIA COVERAGE AND IPO UNDERPRICING The following table 2-2 discuss the literature review focusing on the empirical research about media coverage and IPO underpricing: TABLE 2-2 EMPIRICAL RESEARCHES ABOUT MEDIA COVERAGE Hypothesis. Model And Conclusion. 立. significantly. Kieschnick. positive. + 𝛽7 𝑁𝐴𝑆𝐷𝐴𝑄 𝑟𝑒𝑡𝑢𝑟𝑛 + 𝛽8 𝐿𝑛(𝐻𝑒𝑎𝑑𝑙𝑖𝑛𝑒𝑠) + 𝛽9 𝑅𝑒𝑣𝑖𝑠𝑖𝑜𝑛 + 𝜀 .. Variable definitions：. sit. IPOs, Cook,. IR = β0 + β1 𝐿𝑛(𝑆𝑎𝑙𝑒𝑠) + 𝛽3 𝑆𝑑𝑚𝑖𝑑 + 𝛽4 𝐹𝑙𝑜𝑎𝑡 + 𝛽5 𝑅𝑎𝑛𝑘𝑖𝑛𝑔 + 𝛽6 𝑉𝑒𝑛𝑡𝑢𝑟𝑒. ‧. observe a. adopted. Nat. Marketing of. ‧ 國. They expect to. Variable to be. 學. 1. On the. 政治大. y. Author. er. al. n. and Van Ness, correlation. io. The dependent variable in each regression is the initial return for an initial public. i n U. v. offering (IPO). Ln(Sales) is the natural logarithm of the firm's net sales for the fiscal 2006. between pre-. Ch. engchi. year no later than its offering date. Sdmid represents the standard deviation of the offer publicity. midpoints of the bid-ask spreads during the first trading day. Float represents the and initial IPO ratio of the number of shares issued in the offering to the number of shares issued returns. and outstanding after the offering. Revision equals the difference between the offer 20. 𝑁𝐴𝑆𝐷𝐴𝑄 𝑟𝑒𝑡𝑢𝑟𝑛 𝑅𝑒𝑣𝑖𝑠𝑖𝑜𝑛.

(30) Model And Conclusion. Variable to be adopted. price and midpoint of the initial filing price relative to the midpoint of the initial. 政治大. filing range. Ranking represents an investment banker's ranking from Loughran and. 立. Ritter (2003), but converted to integer rankings. Venture is a dummy variable that. 學. takes on the value one if the IPO firm was backed by a venture capital firm. Nasdaq return is the Nasdaq return over the 15 days prior to the offer date. Ln(Headlines). ‧. io. Conclusion:. sit. Nat. one within six months prior to the offer date.. y. equals the natural logarithm of the number of headlines with the company name plus. er. Hypothesis. ‧ 國. Author. al. n. iv n C Initial returns are positively with pre-offer publicity. Thus, the regular h ecorrelated i U ngch customers of investment bankers clearly benefit from promotional efforts.. 21.

(31) Author. Hypothesis. Model And Conclusion. Variable to be adopted. production. IPO. theories. more media. and Zhang ,. coverage. 2009. relates to. 10 𝑅𝐴𝑁𝐾. + 𝛽11 𝑙𝑜𝑔(𝐴𝑆𝑆𝐸𝑇) + 𝛽12 𝑙𝑜𝑔(1 + 𝐴𝐺𝐸) + 𝛽13 𝑙𝑜𝑔(𝑂𝐹𝐹𝑆𝐼𝑍𝐸). + 𝛽18 90_𝐷 + 𝛽19 𝐵𝑈𝐵𝐵𝐿𝐸_𝐷 + 𝛽20 𝑃𝑂𝑆𝑇𝐵𝑈𝐵𝐵𝐿𝐸_𝐷 + 𝜀 .. Variable definitions：. n. al. er. io. HITS is the number of media articles covering the IPO firm from one day after the greater. HITS OFFERSIZE PREV_D. + 𝛽14 𝑇𝐸𝐶𝐻𝐼𝑁𝑇 + 𝛽15𝑉𝐸𝑁𝑇 + 𝛽16 𝐺𝐿𝑂𝐵𝐴𝐿 + 𝛽17 𝑂𝑉𝐸𝑅𝐻𝐴𝑁𝐺. ‧. Liu, Sherman. 9. 學. predict that. 8. Nat. Underpricing,. 7. y. Coverage and. 治政大 + +𝛽 𝐼𝑁𝐷𝑅𝐸𝑇 + 𝛽 + +𝛽 𝑀𝐾𝑇𝑅𝐸𝑇 + 𝛽 𝐼𝑁𝐷𝑅𝐸𝑇 立. IR = β0 + β1 𝐻𝐼𝑇𝑆 ∗ 𝑃𝑅𝐸𝑉_𝐷 + 𝛽3 𝐻𝐼𝑇𝑆 + 𝛽4 ΔP + +𝛽5 ΔP + 𝛽6 𝑀𝐾𝑇𝑅𝐸𝑇. sit. Information. ‧ 國. 2. Media. i n U. v. filing date to one day before the offer date and then standardizing into per month. Ch. engchi. underpricing. method.. when investor. △p is the percentage price revision, (offer price – midpoint of initial filing. demand is. range)/midpoint of initial filing range.△p+ equals △P when △P is positive and. relatively. zero otherwise. PREV_D equals one when △P is positive and zero otherwise.. 22. MKTRET.

(32) Hypothesis. Model And Conclusion. Variable to be adopted. high, while. MKTRET is value-weighted average daily return for 15 trading days prior to. any. the IPO day. relationship. INDRET is equal-weighted average daily return of firms in the same. between. industry for 15 trading days prior to the IPO day, there industry classification is by. media. Fama-French (1995) 39 industries. INDRET+ equals INDRET when INDRET is. 學. ‧. positive and zero otherwise.. Nat. y. coverage and. 立. 政治大. ‧ 國. is unlikely to. of issuer at IPO. OFFERSIZE is size of the offering, measured as offer price. hold for low-. multiply share offered. TECHINT is equals to 1 if the firm is a technology firm or an. demand. internet firm. VENT equals to 1 if the firm is venture capitalists backed and 0. offerings.. otherwise. GLOBAL equals to 1 if the offering is a global offering and 0 otherwise.. sit. RANK is the ranking of lead underwriter. ASSET is total assets pre-IPO. AGE is age. io. underpricing. n. al. er. Author. Ch. engchi. i n U. v. OVERHANG equals (Pre-IPO shares-secondary shares offering) /(total share. 23.

(33) Model And Conclusion. Variable to be adopted. offering).. 政治大. 90_D equals one if the offering date falls between 1990 and 1998, and. 立. zero otherwise. BUBBLE_D equals one if the offering date falls between 1999 and. 學. 2000, and zero otherwise. POSTBUBBLE_D equals one if the offering date falls between 2001 and 2004, and zero otherwise.. ‧. Conclusion:. y. Nat. io. sit. The relationship is asymmetrical, with positive media coverage associated with more underpricing while no such relation exists for negative media coverage. The relation. n. al. er. Hypothesis. ‧ 國. Author. Ch. i n U. is both statistically and economically significant.. engchi. v. They show that the positive relation between positive media coverage and underpricing is stronger when ex ante uncertainty is greater.. 24.

(34) 2.7 SENTIMENT ANALYSIS To easily define sentiment analysis, this study quotes from the Nasukawa and Yi (2003) entitled their paper, “Sentiment analysis: Capturing favorability using natural language processing”. In the same year, Yi et al. (2003) also release “Sentiment Analyzer: Extracting sentiments about a given topic using natural language processing techniques.” These events together may explain the popularity of “sentiment analysis” among communities self-identified as focused on NLP. A large number of papers mentioning “sentiment analysis” focus on the specific application of classifying. 政治大 caused some authors to suggest that the phrase refers specifically to this narrowly 立 reviews as to their polarity (either positive or negative), a fact that appears to have. ‧ 國. 學. defined task. However, nowadays many construe the term more broadly to mean the computational treatment of opinion, sentiment, and subjectivity in text. Thus, when. ‧. broad interpretations are applied, “sentiment analysis” and “opinion mining” denote the. sit. y. Nat. same field of study.. n. al. er. io. Henry (2008) quantifies tone using computer-based analytical tools to measure the. i n U. v. frequency of positive and negative words found in earnings press releases and. Ch. engchi. concludes the tone of earnings press releases influences investors’ reaction to earnings. Piger and Sedor (2006) find a significant positive association between levels of optimistic tone in earnings press releases and both future return-on-assets and the initial market response. From Tetlock (2007), they conclude that high values of media pessimism induce downward pressure on market prices; unusually high or low values of pessimism lead to temporarily high market trading volume. Furthermore, the price impact of pessimism appears especially large and slow to reverse itself in small stocks. This is consistent. 25.

(35) with sentiment theories under the assumption that media content is linked to the behavior of individual investors, who own a disproportionate fraction of small stocks.. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. 26. i n U. v.

(36) 3. SENTIMENT ANALYSIS MODEL 3.1HYPOTHESES DEVELOPMENT In this section, hypotheses will be discussed and provided as follows： 3.1.1 THE RELATIONSHIP BETWEEN IPO UNDERPRICING AND THE SENTIMENT OF MEDIA COVERAGE Cook et al. (2006) argue that initial IPO returns are positively correlated with pre-issue publicity. There are more studies that had much attention on media coverage, which content analysis of the most important information becomes more. 政治大 offerings is positively related 立 to underpricing. Liu et al. (2009) find the result that the relevant. Thus, Liu et al. (2009) propose that the role of the media in initial public. ‧ 國. 學. positive relationship between media coverage and underpricing is stronger when ex ante uncertainty is greater.. ‧. Sherman and Titman’s (2002) information production theory, when. sit. y. Nat. information is costly, the level of underpricing is determined by the desire for. n. al. er. io. information. For firms with the most to gain from accurate pricing, more investors. i n U. v. will be invited to participate in the offering and underpricing will be greater.. Ch. engchi. Cook et al. (2006) find that underwriters have an incentive to promote an IPO to induce sentiment investors into the market for it. Underwriters are intend to underprice more and the issuer and the underwriters gain wealth for a subsequent seasoned equity offering. It also allows IPO firms to release news that can generate favorable views at their discretion. For instance, a firm can announce its potential operation and business strategies via news. These news shows that the firms have an attempt to present their positive image and attract investors. Therefore, examining financial news of IPO firms in Taiwan allows us to investigate whether the investors are 27.

(37) influenced the positive sentiment of the media. This study therefore forms the following hypothesis: (H1) The positive sentiment of media coverage is positively related to IPO underpricing (initial return). 3.1.2 THE RELATIONSHIP BETWEEN IPO UNDERPRICING AND THE NUMBER OF POSITIVE (NEGATIVE) NEWS REPORTS This study also examines that the number of news items. This approach is with the idea that the number of media is a signal that high media attention is more positive than low media attention. IPOs receive a media signal: companies that have failed to attract. 政治大 are also unlikely to excite investors. 立. media have a “bad” signal, since their inability to excite journalists indicates that they. ‧ 國. 學. In order to research on topic of this study, we still test on the sentiment of media, this study collects the news items from the largest internet news database and categorize. ‧. them (positive, negative, or neutral). Thus, this hypothesis development follows up the. sit. y. Nat. H1 that positive sentiment of media coverage is positively related to IPO underpricing. n. al. er. io. (initial return) and asks whether media coverage is more positive following an initial. i n U. v. return increase and more negative following an initial return decrease.. Ch. engchi. In this hypothesis, this study uses the number of positive (negative) news items to be another proxy as positive (negative) media coverage. There is also a problem to measure the media coverage across firm, since the length of the period varies across firm, this study has to standardize the media coverage measure into per month measure. Therefore, this study constructs this media coverage measure using the one month window before the issue day. This study therefore forms the following hypothesis: (H2) The number of positive (negative) news positively (negatively) relates to underpricing.. 28.

(38) In Figure 3-1, this study shows the framework of the hypotheses, as follows: FIGURE 3-1 HYPOTHESES STUCTURE. Media Coverage. IPO Underpricing H1. •Positive Sentiment of media. •IR. •The number of positive (negative) news H2. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. 29. i n U. v.

(39) 3.2 DATA COLLECTION 3.2.1 SAMPLE SELECTION This study starts out sample construction from January 2010 to December 2014 and collect the IPOs firms which successfully listed on Taiwan Stock Exchange (TWSE) 1. or the over-the-counter GreTai Securities Market (GTSM). for the first time.. Excluding IPOs of financial firms, Taiwan depositary receipts (TDR). This study identifies and extract the sample companies from the list of the Taiwan Economic Journal database (TEJ), and the number of final sample IPOs firms listed on TWSE and. 政治大 of media coverage and two firms that changed their company name. This give us 241 立. GTSM are 75 and 166, respectively. This study then removes one firm due to its lack. sample IPOs. All the public offering data, financial and return data are also obtained. ‧ 國. 學. from the TEJ database.. ‧. The news resources are collected from the Knowledge Management Winner. sit. y. Nat. (KMW) database that is marketed and maintained by the largest financial news group. io. er. in Taiwan. The KMW is the largest news database composed of China Times, China Times Express and Commercial Times. This study searches the news by the company. al. n. v i n Cahtotal of 5,876 newsUitems. This study doesn’t limit abbreviation name, and there are engchi our news articles only to those news items in which is mentioned in the headline or in. the lead paragraph, and this study keeps news sample items contained multiple companies. Because of the characteristic of the financial news and based our analysis on the media coverage, this study does not exclude a large volume of news that actually cover the firm. The search time window is from the day after listing on Emerging Stock Market to the day prior to the issue date of the IPOs (the pre-IPO period), and this study. 1. The over-the-counter GreTai Securities Market (GTSM) officially changed the English title as Taipei Exchange（TPEx）on February 24, 2015. 30.

(40) acknowledges the pre-IPO companies are tend to have more attraction and positive image to the investors after listing on the Emerging Stock Market. There are two other reason that this study uses another time window, one month prior to the issue day. First, since the length of the period varies across firm, this study also standardizes the media coverage measure into per month measure and form another hypothesis to research the relation between the media coverage and underpricing under the same time window. Second, this study follows Liu et.al (2009), they construct the media coverage using the one month before the issue day. Therefore, this study. 政治大. constructs another media coverage measure using the one month window before the issue day.. 立. 3.2.2 SAMPLE DESCRIPTION. ‧ 國. 學. To show the outcome of 241 firms in this study, this study uses the sample. sit. y. Nat. 3-1.. ‧. distribution chart to show the sample characteristic by year and industry given in Table. io. er. Table 3-1 shows that the number of firms is almost evenly distributed in these five years, but 2011 (66 firms) has a large increase in IPO firms. The number of firms is. al. n. v i n concentrated from 2011 to 2013,Cthe proportion of firms h e n g c h i Uin these three years is 68.05%. The breakdown of the 241 IPOs exhibits an unevenly distributed pattern across 24. industries classified by the TSE industrial classification code. In particular, 41 IPOs come from biotechnology & medical care industry, 31 from optoelectronic industry, 29 from the electronic parts & components industry, 24 from semiconductor, 16 from the communications and internet industry, 11 from electronics and peripheral equipment industry, 10 from the information service industry, 10 from the tourism industry and the rest scatter over the remaining 14 industries with a maximum of 9 IPOs and a minimum of 1. As expected, electronics industry related firms dominate the Taiwanese IPO market with 60.17% of the total sample (based on the two-digit Standard Industrial 31.

(41) Classification Code). The distribution of these firms reflects the importance of the electronics products sector, which has become the main driver of the Taiwanese economy for a long time. Obviously, biotechnology & medical Care (41 firms, 17.01%) is blooming up and promoted in Taiwan, becoming the star industry in these years.. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. 32. i n U. v.

(42) TABLE 3-1 SAMPLE DISTRIBUTION Year. 7 3. 3 5. 3 5. 31 29. 12.86% 12.03%. 4 2. 7 6. 4 4. 3 2. 6 2. 24 16. 9.96% 6.64%. 2. 6. 1. 5. 2. 16. 6.64%. 2 4治 4 4 政 2 2 4 大2. 14 11. 5.81% 4.56%. 10 10 9 5. 4.15% 4.15% 3.73% 2.07%. 4. 1.66%. . 1. iv. 3 3. 1.24% 1.24% 1.24% 0.83%. . . 3 .. 4 2 3 1. 2 5 2 2. 1 3 . .. 3 . 1 2. 1. 1. .. 1. 1. .. 2 .. . .. 1 2. er. n. al .. Ch. 1 1. n U engchi . 1 . 1. ‧. ‧ 國. 立. io. Electrical and Cable Iron and Steel Automobile Glass and Ceramics Foods Electronic Products Distribution Total. 9 10. . 1. Nat. Shipping and Transportation Chemical Cultural and Creative Industry Textiles Building Material and Construction Plastics. 9 6. 學. Communications and Internet Electric Machinery Electronics and Peripheral Equipment Information Service Tourism Other Electronic Trading and Consumers' Goods. Percentage 17.01%. y. 2010 2011 2012 2013 2014 Total 5 9 11 11 5 41. sit. Industry Biotechnology & Medical Care Optoelectronic Electronic Parts & Components Semiconductor Other. .. 1. .. .. 3 2. .. .. 1. .. 1. 2. 0.83%. 1 1 . . . .. . . . . 1 1. . . 1 . . .. 1 . . 1 . .. . 1 . . . .. 2 2 1 1 1 1. 0.83% 0.83% 0.41% 0.41% 0.41% 0.41%. 37. 66. 51. 47. 40. 241. 100%. 33.

(43) TABLE 3-2 NEWS DISTRIBUTION- BY INDUSTRY AND NEWS CLASSIFICATION the pre-IPO period Industry. one month prior to. News Classification. the issue day The. Percenta The. Perce. number. ge. ntage. number. of news All. 5876. Positive news. 立. Neutral news. 1498. 93.28%. 48. 2.99%. 205. 3.49%. 60. 3.74%. All. 3297. 100%. 844. 100%. Positive news. 2974. 90.20%. Neutral news. 189. 5.73%. Negative news. 134. 4.06%. 1347. 100%. io. Positive news. n. & Medical Care. Ch. y. Nat. al. All. sit. ‧ 國. Negative news. Biotechnology. Other Industry. 100%. ‧. Industry. 治 91.08% 政 5352 大 319 5.43%. 1606. 學. Electronics. 100%. er. All Sample. of news. v ni. e n g1265 c h i U93.91%. 778. 92.18%. 28. 3.32%. 38. 4.50%. 335. 100%. 320. 95.52%. Neutral news. 49. 3.64%. 4. 1.19%. Negative news. 33. 2.45%. 11. 3.28%. All. 1232. 100%. 427. 100%. Positive news. 1113. 90.34%. 400. 93.68%. Neutral news. 81. 6.57%. 16. 3.75%. Negative news. 38. 3.08%. 11. 2.58%. 34.

(44) Since the length of the period varies across firm, this study also standardizes the media coverage measure into per month measure to compare to the quantity of news across from the IPO firms. Therefore, this study also counts the number of the news prior to the IPO issue date. Table 3-2 shows that the characteristics of 5876 news items collected on the pre-IPO period. This study divides the news into three types： 1) positive news, 2) neutral news 3) negative news. News that is considered to have a positive (negative) impact on IPO firms. News is considered to have a trivial impact is classified as neutral news. Table 3-2 presents a majority of news have a positive. 政治大 proportion of positive news accounts for 93.28% one month before the IPO issue 立. impact on the IPO firms in that positive news show up 91.08%. Moreover, the. date. It is not surprise that all of the three industries has a higher percentage of. ‧ 國. 學. positive news. As expected, electronics industry related firms cover with 56.11%2. ‧. of the total news sample.. sit. y. Nat. 3.3 REASERCH METHOD. io. NEWS ITEM. er. 3.3.1 APPROACHES TO MEASURE THE SENTIMENT POLARITY IN. al. n. v i n C hinto one of three categories: This study classifies each news “positive”, “negative” engchi U. or “neutral.” The classification is based on dictionary-based approach to convert the qualitative news to quantitative measures. In the past studies, using a dictionary to. compile sentiment words is an obvious approach because most dictionaries (e.g., WordNet Miller et al., 1990) list synonyms and antonyms for each word. Thus, a simple technique in this approach is to use a few seed sentiment words to bootstrap based on the synonym and antonym structure of a dictionary. Specifically, this method works as follows: A small set of sentiment words (seeds) with known positive or negative. 2. 3297/5876=56.11% 35.

(45) orientations is first collected manually, which is very easy. The algorithm then grows this set by searching in the WordNet or another online dictionary for their synonyms and antonyms. This study also motivated by these studies. Hence, there are three graduate students read the financial news from KMW news database, and this study extracts the most often sentiment word written by the journalists. This study specifies a dictionary which includes the positive and negative sentimental words. However, in this study, this study doesn’t extend our sentiment dictionary by using the dictionaries list synonyms and antonyms for each word. The way of building sentimental lexicons. 政治大 words from the news items, this study finds that the number of the positive (negative) 立. is based on Lin and Chen (2013). In the process of manually extracting the sentiment. word is more than negative (positive) word in the positive (negative) news items.. ‧ 國. 學. Therefore, to avoid the subjective judgment bias, the computer program scan over the. sit. y. Nat. dictionary.. ‧. texts of news and calculate the number of the sentimental words in our sentiment. io. er. In this study, we also use sentiment dictionary-based to classify the news items into positive, neutral, or negative news. In Figure 3-2, this study shows the approach of. al. n. v i n C h the number of positive news sentiment classification. When words is more than the engchi U. number of negative words in each news item, the news item will be classify into positive news. When the number of positive words equal to the number of negative words in each news item and the number of negative words is less than six, the news item will be classify into neutral news. When the number of negative words is more than the number of positive words in each news item and the number of negative words is more than six, the news item will be classify into negative news. Where there are more negative words in news item, it expresses bad signal to the investors. Comparing to the classification by manually and to have a more accurate classification of news, that’s why this study sets a threshold of six negative words to do classification. 36.

(46) FIGURE 3-2 THE APPROACH OF NEWS SENTIMENT CLASSIFICATION. 政治大 3.3.2 LYDIA SENTIMENT ANALYSIS 立. How Lydia sentiment analysis works can be found from (Bautin, Vijayarenu, and. ‧ 國. 學. Skiena 2008; Godbole, Srinivasaiah, and Skiena 2007). The Lydia sentiment data. ‧. consists of time series of favorable (positive) and unfavorable (negative) words co-. sit. y. Nat. referenced with occurrences of each named entity (here denoting companies). Let p and. io. times in the corpus (including neutral references).. n. al. Ch. Then it can give below derived measures: • polarity = (p − n)/(p + n). engchi. er. n denote the number of raw positive and negative references, which occurs a total of N. i n U. v. • subjectivity = (n + p)/N Polarity indicates percentage of positive sentiment references among total sentiment references, while subjectivity indicates proportion of sentiment to frequency of occurrence. These derived measures could provide additional information that raw references cannot.. 37.

(47) 3.4 REGRESSION MODEL As mentioned in Section 3.2.1. The search window is pre-IPO period, and also construct another standard time window, one month before the issue day. As shown in Table 3-3, therefore, this study forms the different models to test news data between the two different periods. This study also tests our hypothesis under different industries. At the industry level, the importance of the electronic products sector, which has become the main driver of the Taiwanese economy for a long time is also a key characteristics of IPO firms. In Taiwan, This study thinks the electronics industry has. 政治大 supply chain bloom up, the rest of the supply chain would benefit from the trend. Hence, 立 the overwhelming advantage on the cluster effect of media. If one of the electronics. this study also researchs on that the positive sentiment of media coverage about the. ‧ 國. 學. companies in the electronics industry positively relates to underpricing than those in. ‧. non-electronics industry.. sit. y. Nat. io. al. The Pre-IPO Period. n. Sample. Ch. H1-Model (1)、 Full Sample. Electronics Industry. er. TABLE 3-3 FRAMEWORK OF THE MODEL. v. One Month Prior to The Issue Day. i n U. e n g c h iH1-Model (2) 、H1-Model (4)、. H1-Model (3). H2-Model (5). H1-Model (1). H1-Model (2)、H2-Model (5). H1-Model (1). H1-Model (2)、H2-Model (5). Non-Electronics Industry. 38.

(48) 𝐈𝐑 = 𝛃𝟎 + 𝛃1 𝐓𝐎𝐍𝐄_𝐑𝐀𝐓𝐈𝐎 + 𝛽2 𝑻𝑳𝑯𝑰𝑻 + 𝛽3 𝑰𝑷𝑶_𝑷𝑬𝑹𝑰𝑶𝑫 + 𝛽4 𝑳𝑵_𝑭𝑰𝑹𝑴𝑺𝑰𝒁𝑬 + 𝛽5 𝑺𝑪𝑨𝑳𝑬𝑫_𝑶𝑭𝑭𝑬𝑹 + 𝛽6 𝑴𝑲𝑻𝑹𝑬𝑻 + 𝜀 .. (1). 𝐈𝐑 = 𝛃𝟎 + 𝛃1 𝐓𝐎𝐍𝐄_𝐑𝐀𝐓𝐈𝐎 + 𝛽2 𝑻𝑳𝑯𝑰𝑻 + 𝛽3 𝑰𝑷𝑶_𝑷𝑬𝑹𝑰𝑶𝑫 + 𝛽4 𝑳𝑵_𝑶𝑭𝑭𝑬𝑹𝑺𝑰𝒁𝑬 + 𝛽5 𝑺𝑪𝑨𝑳𝑬𝑫_𝑶𝑭𝑭𝑬𝑹 + 𝛽6 𝑴𝑲𝑻𝑹𝑬𝑻 + 𝜀 . (2) Equation (1) and (2) is to test hypotheses (H1) which investigate the association. 政治大 study uses control variable LN_FIRMSIZE in Equation (1) and (3), but substitute the 立 between underpricing and the sentiment of media before IPO issue date, and this. LN_FIRMSIZE for LN_OFFERSIZE in Equation (2) and (4). There are two reasons. ‧ 國. 學. to explain why this study uses different model to test H1. First, the LN_FIRMSIZE. ‧. and LN_OFFERSIZE have a highly positive association, and it is more reasonable. sit. y. Nat. to use LN_OFFERSIZE control variable when test the samples of one month prior. io. er. to the issue day based on the timing of the pre-IPO period. Second, when this study substitutes the LN_FIRMSIZE for LN_OFFERSIZE in Equation (2), the adjusted R-. n. al. square is higher.. Ch. engchi. i n U. v. 𝐈𝐑 = 𝛃𝟎 + 𝛃1 𝐓𝐎𝐍𝐄 + 𝛽2 𝑻𝑳𝑯𝑰𝑻 + 𝛽3 𝑰𝑷𝑶_𝑷𝑬𝑹𝑰𝑶𝑫 + 𝛽4 𝑳𝑵_𝑭𝑰𝑹𝑴𝑺𝑰𝒁𝑬 + 𝛽5 𝑺𝑪𝑨𝑳𝑬𝑫_𝑶𝑭𝑭𝑬𝑹 + 𝛽6 𝑴𝑲𝑻𝑹𝑬𝑻 + 𝜀 .. 𝐈𝐑 = 𝛃𝟎 + 𝛃1 𝐓𝐎𝐍𝐄 + 𝛽2 𝑻𝑳𝑯𝑰𝑻 + 𝛽3 𝑰𝑷𝑶_𝑷𝑬𝑹𝑰𝑶𝑫 + 𝛽4 𝑳𝑵_𝑶𝑭𝑭𝑬𝑹𝑺𝑰𝒁𝑬. (3) ). + 𝛽5 𝑺𝑪𝑨𝑳𝑬𝑫_𝑶𝑭𝑭𝑬𝑹 + 𝛽6 𝑴𝑲𝑻𝑹𝑬𝑻 + 𝜀 . (4) ) With earlier literature suggesting the positive relationship between media coverage and underpricing is stronger when ex ante uncertainty is greater, and fail to 39.

(49) find any relationship between positive media coverage and IPO firms’ long run under-performance (Liu et.al 2009). This study uses two approaches to confirm whether the IPO underpricing reflects the public media sentiment, one approach is using the independent variable of TONE_RATIO, the other is using the independent variable of TONE. There independent variables TONE_RATIO and TONE are using dictionary-based approach to measure sentiment extent of the media. This dictionarybased approach may lead to a result that this study doesn’t view each news item as the same weight to measure the sentiment extent about the target firm, however when. 政治大 The difference these two independent variable is TONE_RATIO scaled by the 立. scoring by calculating the frequency of the words means weighted average method.. number of the dictionary words to standardize the sentiment score. When scaled by. ‧ 國. 學. the number of the dictionary words, the weighted average is more apparent. Thus. er. io. sit. y. Nat. TONE.. ‧. TONE_RATIO emphasizes the different weight score of each news items more than. 𝐈𝐑 = 𝛃𝟎 + 𝛃1 𝐍𝐔𝐌_𝐏𝐎𝐒 + 𝛽2 𝑵𝑼𝑴_𝑵𝑬𝑮 + 𝛽3 𝑰𝑷𝑶_𝑷𝑬𝑹𝑰𝑶𝑫. al. n. v i n C h + +𝛽5𝑺𝑪𝑨𝑳𝑬𝑫_𝑶𝑭𝑭𝑬𝑹 + 𝛽4 𝑳𝑵_𝑶𝑭𝑭𝑬𝑹𝑺𝑰𝒁𝑬 + 𝛽6 𝑴𝑲𝑻𝑹𝑬𝑻 + 𝜀 . engchi U. (5) Although Equation (5) is to test the hypotheses (H2) and also to find the association between underpricing and the number of positive (negative) news before IPO issue date, but the number of positive or negative news means the media coverage and media visibility of the target firm. There independent variables NUM_POS and NUM_NEG are using dictionary-based approach to classify the news mentioned in section 3.3.1, and there is no concept of weighted average method.. 40.

(50) Where: TONE_RATIO = the score of the frequency count of positive words divided by the positive sentiment dictionary minus the ratio of the frequency count of negative words divided by the negative sentiment dictionary. TONE = the score calculated as (Positive word-Negative word) divided by (Positive word+ Negative word). TLHIT = the total words count excluding symbol words in each news item using computer program.. 政治大 during the period from the one month prior to the issue day. 立. NUM_POS = the number of positive news item about the firm i published. NUM_NEG = the number of negative news item about the firm i published. ‧ 國. 學. during the period from the one month prior to the issue day.. ‧. LN_FIRMSIZE = the natural logarithm of the market value of the IPO firm at. sit. y. Nat. the offering and is noted in millions of New Taiwan dollars.. io. er. LN_OFFERSIZE = the natural logarithm of the IPO issue size, measured as offer price multiplied by the number of shares offered.. al. n. v i n C h logarithm of difference IPO_PERIOD = the natural in the number of days engchi U between the date of listing on Emerging Stock Market and the issue date.. SCALED_OFFER = the dummy variable equals one if revision is greater than one, and zero otherwise. MKTRET = the market return for 30 trading days prior to issue day, in percent. This study uses the market data called from TEJ database 3.4.1 DEPENDENT VARIABLE This study begins our analysis on IPO underpricing, also called IPO firms’ initial returns. Because there is no lock-up period for an IPO in Taiwan for our sample period. Hence, this study takes the closing price of issue date as the first day market price of 41.