Pre-IPO Market, Underwritting Procedure, and IPO Performance

Transcription

1 Pre-IPO Market, Underwritting Procedure, and IPO Performance Hsuan-Chi Chen a, Sue-Jane Chiang b*, Pei-Gi Shu b a Anderson School of Management, University of New Mexico, Albuquerque, NM 87131, USA b Department of Business Administration, Fu-Jen Catholic University, 242 HsinChuang, Taiwan * Corresponding author Draft: January 15, 2018 Abstract Using a unique data of mandatory pre-ipo market trading in Taiwan, we examine the joint effects of pre-ipo market trading and underwriting procedures on the short-term and long-term IPO performance. The empirical results of 476 IPOs in the sample period suggest that for the fixed price public offers, on average, the offer price is set close to the pre-ipo market price, which is subject to insider manipulation and significantly higher than the middle point of the price range set by the bookbuilding offers. The higher (lower) offer price is followed by lower (higher) initial return and poorer (better) long-run performance. The self-dealing hypothesis relevant to issuers, investors, and underwriters can help explain the empirical findings. Keywords: Pre-IPO market, Bookbuilding, Fixed price offer 1

2 1. Introduction The high average initial return or IPO underpricing, the percentage change from the offer price to the first-day closing price, has been widely documented in the IPO literature. Intuitively, the IPO underpricing anomaly could be attributed to the underestimation of the offer price, the overestimation of the initial closing price in the aftermarket, or both. Several theories have been proposed to shed light on relatively low offer prices, including winner s curse (Rock,1986), signaling theory (Allen and Faulhaber, 1989), information asymmetry (Ritter, 1984; Ritter and Welch, 2002), and agency problems (e.g., Ljungqvist, 2007; Ritter, 2011). 1 In general, agency problems of underwriters or the conflict of interests are important in explaining IPO underpricing (e.g., Chang et al., 2016). As for the overestimation of the initial closing prices in the IPO aftermarket, investor overvaluation, overoptimistm, or overreaction may be the main factors. If the high initial returns are primarily attributed to the relatively low offer prices, the fair initial closing prices in the aftermarket are less likely to lead to poor post-ipo long-run performance. In contrast, if the high initial returns are mainly attributed to the overestimation of initial closing prices, the higher initial returns would be associated with long-run underperformance due to market correction. No matter which scenario would prevail, undervaluation of offer price or overvaluation of the initial closing price, the crucial nexus that results in the abnormal initial return relies on the malfunction of price discovery associated with the IPO market. How to enhance price discovery for IPO shares has been explored by academics and practitioners. The common practices of obtaining pricing-relevant information directly from potential investors so as to enhance price discovery include bookbuilding (e.g., Benveniste and Spindt, 1989; Spatt and Srivastava, 1991; Sherman, 2005), auction (e.g., Biais and Faugeron-Crouzet, 2002; Derrien and Womack, 2003), and pre-ipo trading (e.g., Derrien and Kecskes, 2007; Chang et al., 2016). 2 However, it is not clear which one would prevail when they are juxtaposed, and whether they are substitutes or complementary in terms of price discovery. Benveniste and Spindt (1989) and Spatt and Srivastava (1991) argue that through the discretion in share allocation underwriters can collect true market 1 Still, there are other plausible explanations, including the ease of marketing and institutional investors obtaining kickback from share placement and trading commission (See Loughran and Ritter, 2002). 2 The use of IPO auction is not popular around the world. For our investigated case of Taiwan, there were 53 IPOs adopting auction prior to the introduction of compulsory pre-ipo market trading. There were only two IPO auctions after the introduction of compulsory pre-ipo market trading during our sample period. We do not include these two IPO firms in our sample. 2

3 information from sophisticated investors to set offer prices that are close to the market prices and then reduce IPO underpricing. Sherman (2005) indicates that bookbuilding allows underwriters to manage investor access to IPO shares, facilitating them to reduce risk for both issuers and investors and to control spending on information acquisition, thereby limiting either IPO underpricing or aftermarket volatility. From the perspective of reducing information asymmetry, bookbuilding should be preferred to fixed price offers. However, self-interested underwriters are strongly motivated to price the offering downwards so that they can allocate underpriced shares to their favored clients in exchange for side payments (Ljungqvist, 2007; Loughran and Ritter, 2002; Reuter, 2006; Ritter, 2011; Nimalendran, Ritter and Zhang, 2007; Liu and Ritter, 2010 and 2011; Goldstein, Irvine and Puckett 2011). If this is the case, traditional bookbuilding IPOs tend to exhibit lower offer prices and therefore higher initial returns than those associated with other offering methods. In addition to underwriting procedures, the existence of pre-ipo market is supposed to facilitate price discovery for IPO shares. However, the empirical findings are mixed. Derrien and Kecskes (2007) posit that the pre-ipo market that provides a two-stage offering may help reduce the valuation uncertainty for IPO firms and lead to relatively low initial returns. In contrast, Cornelli, Goldreich, and Ljungqvist (2006) and Dorn (2009) indicate that individual investors who are likely to be noise traders tend to trade in the pre-ipo market. The prices of IPO shares are affected by investor sentiment that results in long-run underperformance and volatile short-run initial returns. If pre-ipo market trading is helpful to facilitate price discovery, the initial returns would be significantly reduced following its introduction, and the reduction in initial returns should be equally applicable to both fixed price offers and bookbuilding offers. By contrast, if underwriting methods rather than the introduction of pre-ipo market matter, then only a certain kind of underwriting procedure has an effect on the reduction of underpricing. Because the bookbuilding and fixed price offers are both available for underwriting IPO shares and pre-ipo market trading is mandatory for firms seeking IPOs in Taiwan for some periods of time, as a natural experiment, we use the IPO sample of Taiwan for the period to examine whether pre-ipo market trading or underwriting procedures would prevail in affecting initial returns. The empirical results for 476 IPOs show that the average initial return of IPO firms with bookbuilding is significantly higher than that of firms adopting fixed price offers. We further examine the long-run wealth relatives for both groups of firms and find - 3 -

4 that firms adopting bookbuilding offers exhibit stronger long-run performance than firms adopting fixed price offers. The results are consistent with the idea that the high initial returns of bookbuilt IPOs are primarily attributed to the relatively low offer prices rather than the overestimation of initial closing prices. We find that in some cases the pre-ipo market prices are even higher than the initial aftermarket prices. 3 This observation motivates us to propose the self-dealing hypothesis to explain the difference in offer prices and therefore initial returns between fixed price offers and bookbuilding offers in the context of mandatory pre-ipo market trading. In fixed-price offers self-dealing insiders would boost the pre-ipo prices to benefit their wealth given that the thin-trading pre-ipo market may faciliatet them to manipulate the pre-ipo market prices. The offer prices set with reference to the pre-ipo market prices tend to be acceptable to uninformed individual investors who are the major subscribers to IPO shares in fixed price offers. By contrast, in bookbuilding offers, there are at least two motives for underwriters to set relatively low offer prices compared to the aftermarket prices. On the one hand, underwriters underprice IPO shares to induce truthful information revelation from institutional investors. On the other hand, self-interested underwriters have strong incentives to set the offer prices downwards so that they can allocate underpriced shares to their favored clients in exchange for side payments (Loughran and Ritter, 2002; Reuter, 2006; Nimalendran, Ritter and Zhang, 2007; Liu and Ritter, 2010 and 2011; Goldstein, Irvine and Puckett 2011). Because the underwriters of bookbuilding offers in Taiwan have larger bargaining power in setting the offer prices, they tend to set offer prices close to the middle of prelimnary offer price range. Moreover, institutional investors participating in the bookbuilding offers are rational and less gullible by the pre-ipo market prices that are subject to insider manipulation. Our empirical results support the self-dealing hypothesis by showing that the offer prices of fixed-price offers set with reference to the pre-ipo market prices are significantly higher than the offer prices of bookbuilding offers set close to the middle of preliminary offer price range. In addition, the theory of information cascade suggests that investors who observe the investment choice made by previous investors can update their beliefs about the value of IPO shares (Welch, 1992). The theory implies that individual investors may update their beliefs based on pre-ipo market prices so that pre-ipo market prices are positively related to aftermarket prices. If offer prices are adjusted 3 Prior studies also find that the pre-ipo market prices are higher than initial aftermarket prices (for example, see Derrien and Kecskes (2007), Cornelli, Goldreich, and Ljungqvist (2006), Aussenegg, Pichler, and Stomper (2006), and Dorn (2009)

5 upwards under the circumstance of artificially boosting pre-marker prices, the short-run initial returns may be lower. However, stock performance may be poor if the initial overvaluation is corrected in the long run. In contrast, if institutional investors and underwriters of bookbuilding offers do not rely on pre-ipo market prices very much and set the offer prices based on their own information and evaluation, information cascade arising from pre-ipo market trades leads to short-run overreaction at most in the aftermarket. Since the offer prices of bookbuilding offers only partly reflect the level of pre-ipo market prices, the short-run initial returns would be higher than those of fixed-price offers. Therefore, the long-run stock performance of bookbuilding offers is expected to be better than that of fixed price offers if the initial overreaction is corrected in the long run. In sum, our arguments predict that the relatively lower (higher) offer prices are associated with higher (lower) initial return and better (worse) long-run performance. Both the introduction of pre-ipo market trading in 2003 and the introduction of bookbuilding procedure in 2004 aim at enhancing price discovery for IPO firms. 4 The natural setting of the Taiwan stock market with a sequential adoption of the mandatory pre-ipo market trading and then bookbuilding offer allows us to investigate which mechanism is effective in addressing IPO anomalies. Moreover, when long-run performance is of interest, which one would prevail in mitigating the alleged long-run underperformance of IPO firms? Our finding suggests that underwriting procedure rather than pre-ipo market trading matters for both price discovery and long-run post-ipo performance. According to the self-dealing hypothesis, the higher the initial return is, the stronger the long-run performance. Our findings have several policy implications. First, no matter what underlying theories applied to explain abnormal initial returns, the most critical point is the biased setting of offer price. 5 How to reduce the information asymmetry or agency problem between issuing firms and outsider investors and enhance price discovery remains the most crucial issue for regulatory entities. Our findings from the natural experiment of the IPO market in Taiwan suggest that bookbuilding rather 4 The Emerging Stock Market was initially instituted in IPO firms had the option of choosing pre-ipo market trading at that time. Since January 1, 2003, all IPO firms had been required to have at least 3-month pre-ipo market trading before the application of IPO. In January 2005, the duration of mandatory pre-ipo trading was revised to 6 months. Furthermore, though bookbuilding offer was initially allowed in 1995, it did not gain popularity because it is restricted to newly issued shares only. At that time, old shares owned by pre-ipo shareholders are put for public subscription. Bookbuilding offer has not been widely adopted by IPO firms until 2004 when the regulation stipulated that only newly issued shares are eligible for public subscription. 5 Previous studies illustrate numerous possible explanations including information asymmetry (e.g., Ritter, 1984; Ritter and Welch, 2002), winner s curse (e.g., Rock, 1986), signaling, timing (e.g., Allen and Faulhaber, 1989), overvaluation, over-optimistic, or over-reaction

6 than pre-ipo market matters in ameliorating biased offer price and therefore enhancing price discovery. Second, our sample allows us to compare bookbuilding offers with fixed-price public offers given the mandatory pre-ipo market trading. Bookbuilding is believed to be an ideal underwriting procedure in terms of price discovery (Benveniste and Spindt, 1989; Sherman, 2000). Sherman (2005) also illustrates the popularity of bookbuilding. It happens to the Taiwan IPO market that bookbuilding had soon become the main stream in 2005 since its initial induction in In general, as compared to fixed-price public offer, bookbuilding is a better underwriting procedure through which underwriters could collect information from informed investors before setting the offer price. 7 Our finding indicates that even with the existence of pre-ipo market, the issues such as information asymmetry, winner s curse, and signaling that confronts the underwriters of fixed price offers were not effectively tackled. Rather, the noise that exists in the pre-ipo market and biases the setting of offer price for a fixed-price public offer remains salient in affecting the aftermarket prices. For fixed-price public offers, setting offer prices lower due to the concerns with information asymmetry, winner s curse, and signaling coupling with higher aftermarket prices arising from individual investors overoptimism or overreaction leads to higher initial returns (e.g. Benveniste and Spindt, 1989; Spatt and Srivastava, 1991; Loughran, Ritter and Rydqvist, 1994). Our finding renders little lenity to the proclaimed value of pre-ipo market trading. Prior studies so far do not converge regarding the value of pre-ipo market trading. Derrien and Kecskes (2007) indicate that the two-stage offering strategy prevailing in the U.K. market is less costly than an IPO because trading reduces the valuation uncertainty of these firms before they issue equity. They find that firms with two-stage offering are associated with 10% to 30% lower in initial returns than for comparable IPOs. By contrast, Goldreich and Ljungqvist (2006) and Dorn (2009) indicate that retail buyers consistently overpay in the when-issued market relative to the immediate aftermarket, and that was mainly due to investor sentiment. Sentiment serves as a driver to affect when-issue prices and poor aftermarket returns. Cornelli, Goldreich, and Ljungqvist (2006) use the midpoint of the filing price range as a proxy of the fundamental value and the difference between fundamental value and 6 The underwriting system launched in 2004 indicates that bookbuilding IPOs need to have 50% of issued shares put for open subscription. However, the underwriting system of having half bookbuilding and half fixed-price offer indeed prolonged the overall underwriting schedule and therefore was censured for inefficiency. In 2005 the underwriting system was reformed to be more flexible that underwriters have the discretion of setting the proportion put for bookbuilding offer and the proportion for fixed price offer which is subject to the ceiling of 20%. 7 Prior studies indicate that for tackling the problem of winner s curse fixed-price offers need to have higher price discount (e.g., Benveniste and Wilhelm, 1990, Spatt and Srivastava, 1991, Benveniste and Busaba, 1997 and Biais and Faugeron-Crouzet, 2002) - 6 -

7 grey market price as a proxy of investor sentiment. They find that in high sentiment (indicating overoptimism) the grey market prices are a very good predictor of first-day aftermarket prices, while they are not in low sentiment. Moreover, long-run price reversal only exists in high sentiment. Our finding lends support to the limited value of pre-ipo market trading that is unable to enhance price discovery. Rather, the pre-ipo prices are subject to the adverse impact of individual investor sentiment and/or manipulation. In addition to the above studies, Chang et al. (2016) is probably the closest one to ours. They examine 218 bookbuilding IPOs in Taiwan from the period and find that the existence of the pre-ipo market facilitates price discovery in the sense that the pre-ipo market price is predictive of the initial aftermarket price. However, it remains a puzzling phenomenon that the average (median) initial return of 55.3% (36.7%) seems to be inconsistent with the alleged benefit of pre-ipo market. Nevertheless, they cite the argument put forth by Ritter (2011) that information asymmetry comprises only a small portion of initial return. Rather, the agency problem that is derived from underwriters rent-seeking behavior is the main driver for a high average initial return. The major difference between our study and Chang et al. (2016) is that we trace the sample back to 2003 and include both bookbuilding and fixed-price public offers. The extension of sample period allows us to make a comparison between the bookbuilding offers and fixed-price offers. Therefore, we can test the self-dealing hypothesis by simultaneously including both underwriting procedures and pre-ipo market. The unique feature of our study is in sharp contrast with prior studies that use bookbuilding IPOs only (e.g., Cornelli, Goldreich, and Ljungqvist, 2006; Aussenegg, Pichler, and Stomper, 2006; Dorn, 2009). Our study potentially makes the following incremental contributions to the IPO literature. First, we find that underwriting procedures but not pre-ipo market trading prevails in affecting initial returns and long-run performance. Second, we propose the self-dealing hypothesis that helps explain the difference in setting offer prices between fixed-price public offers and bookbuilding offers. Third, even though our finding of the positive relation between pre-ipo market prices and aftermarket prices is consistent with the theory of information cascade (e.g. Derrien and Kecskes, 2007; Goldreich and Ljungqvist, 2006; Dorn, 2009; Chang et al., 2016), we are not as optimistic as prior studies that support the efficient pre-ipo market trading. For example, Derrien and Kecskes (2007), Goldreich and Ljungqvist (2006) and Dorn (2009) indicate that pre-ipo market prices are a very good predictor of first-day aftermarket price. Chang et al. (2016) argue that pre-ipo market prices are very informative about post-market prices. We find that pre-ipo market trading is subject to insider manipulation and noisy trading (e.g

8 Cornelli, Goldreich, and Ljungqvist, 2006; Dorn, 2009) and the positive relation between pre-ipo market prices and aftermarket prices may be attributed to information cascade. Finally, we connect the underwriting mechanism, initial return, and long-run performance and find that underwriting mechanism is a major determinant if offer prices fully (partially) reflect the levels of pre-ipo market prices, then the initial return is relatively low (high) and long-run performance is relatively weak (strong). Specifically, for fixed-price public offers, pre-ipo market trading facilitates to keep both the offer price and aftermarket price relatively high, followed by relatively lower average initial return of 8.95% and worse long-run stock performance. In contrast, for bookbuilding offers, the offer prices are set relatively low, which is then followed by relatively higher average initial return of 44.92% and better long-run stock performance. The rest of this paper proceeds as follows. Section 2 describes Taiwan s IPO Market. Section 3 reviews the relevant literature and develop hypotheses accordingly. Section 4 discusses the data, variables, and empirical models. Section 5 presents the empirical results. Section 6 concludes the study. 2. IPO Market in Taiwan As compared to IPO markets in other countries, the IPO market in Taiwan is associated with two special attributes. First, since 2003 all IPO firms in Taiwan have been required to have at least three-month pre-ipo market trading before listing. Unlike other IPO markets where pre-ipo market trading is optional, the mandatory pre-ipo market trading in Taiwan is exceptional and provides a clean IPO sample without selection bias to examine the effectiveness of pre-ipo market trading. Second, there are alternative underwriting procedures including bookbuilding, fixed-price public offers, and auction to be freely chosen by issuing firms. This feature allows us to compare the effectiveness between underwriting procedures. 2.1 Pre-IPO market The main purpose of the pre-ipo market in Taiwan (the Emerging Stock Market, ESM) instituted in 2002 was to enhance information transparency and visibility for both firms preparing for getting listed following their IPOs and firms delisted from major secondary markets. At that time, firms preparing for their IPOs had the option to decide whether their shares are traded in the pre-ipo market before getting listed formally. Starting January 1, 2003, all IPO firms were required to have at least 3-month pre-ipo market trading before getting listed. Since 2005, it was mandatory for at least 6-month pre-ipo market trading before get listed on either the Taiwan - 8 -

9 Stock Exchange (TWSE) or Gre-Tai Securities Market (GTSM), the two major stock markets in Taiwan. Compared to firms listed on TWSE and GTSM, firms that trade in the pre-ipo market are subject to less rigorous requirements: they need to disclose financials (audited annual/semi-annual financial statement) and important corporate events. Moreover, they need to have written recommendations by two or more securities firms, with one of which being the lead underwriter in the latter IPOs. The trading in ESM is facilitated by market makers who come from the recommending securities firms and offer bid and ask quotes during normal trading hours through the Emerging Stock Computerized Price Negotiation and Click System (the Click System). This is different from the order-driven system applied in TWSE and GTSM markets. The ESM is a dealer market in which the recommending securities firms act as market makers to assume the responsibility for continuous trading. They quote bid and ask prices through the click system with the bid-ask spread not exceeding 7% of the ask price. The market makers have the obligation to trade at the quoted price when the order is smaller than 2,000 shares. For large order that exceeds 10,000 shares, the trade is conducted through direct negotiation by letter, telephone, or face-to-face talk. This is in sharp contrast with TWSE and GTSM markets where all trades are through fully automated electronic trading systems and only limit orders are accepted. Moreover, unlike the price limit applied in TWSE and GTSM, there is no price limit in the ESM. 8 Though both individual and institutional investors can trade on the ESM, institutional investors are less likely to invest in ESM stocks because of high risks associated with these stocks. 9 Therefore, ESM trading is mainly comprised of retail investors who tend to be noise traders. Moreover, insiders such as directors, supervisors, and shareholders with more than 10% ownership are restricted from selling after the firm has applied for an IPO. To sum up, several attributes associated with the ESM in Taiwan including the quote system, no price limit, the main composition of individual investors, and high liquidity risk give rise to price manipulation in the ESM. 10 The ESM in Taiwan is similar to U.K. s Alternative Investment Market (AIM) in which firms preparing for listing have the option to trade before getting listed. 8 Since June 1, 2015, the price limit has been 10% instead of 7% that had been applied in TSWE and GESM for a long period of time. 9 Since 2011 mutual funds have been allowed to invest in the ESM. 10 The prices are easily manipulated because of low daily trading volume and no price limit. Moreover, the high prices that are likely subject to insider manipulation are very attractive to individual investors who tend to herd in trading stocks

10 Nevertheless, there are several differences between the ESM in Taiwan and the AIM in the U.K. First, pre-trading is mandatory for Taiwan s IPO firms while optional for U.K. IPO firms. Derrien and Kesckes (2007) find that there were only 11% of IPO firms for the period choosing the AIM for pre-market trading. Moreover, there is no minimum duration for pre-ipo market trading for U.K. IPO firms. But Taiwan s IPO firms are required to be traded in the ESM for at least 6 months before getting listed. Another similar type of pre-ipo market is found in the when-issued markets prevailing in the European countries. Again, these pre-ipo markets are not mandatory for European IPO firms and most of them are not unified and vary widely from country to country (Cornelli, Goldreich and Ljungqvist, 2006). The pre-ipo market trading in the European countries lasts only for a week or so before an IPO firm gets listed. In a nutshell, there are three distinctive features associated with Taiwan s ESM in our sampling period: (1) it is mandatory for firms seeking IPOs; (2) the duration of pre-ipo trading is at least 3 (or 6) months; (3) the coexistence of pre-ipo market trading and alternative underwriting procedures: bookbuilding and fixed-price public offer. These distinctive features allow us to conduct a thorough investigation into the impact of pre-ipo market on subsequent IPO stock performance. Specifically, mandatory pre-ipo trading is free from the concern of selection bias. The long duration of pre-ipo trading allows us to get a clear picture of how pre-ipo trading affects subsequent prices in the offering and its aftermarket. The coexistence of pre-ipo trading and alternative underwriting procedures gives us an opportunity to examine whether pre-ipo market or underwriting procedures matter for the enhancement of price discovery. 2.2 IPO underwriting methods Fixed-price public offer was the only underwriting procedure in early days. To enhance the marketability of newly issued shares, both the bookbuilding and auction methods were inducted in Nevertheless, most bookbuilding IPOs in Taiwan adopt hybrid bookbuilding in which half of the newly issued shared are reserved for the tranche of fixed-price public offer. 11 However, the bookbuilding method at its first induction in Taiwan could be only applied to the issuance of primary shares. This is in sharp contrast with the preference of issuing firms for using secondary shares in IPOs. One reason for such a preference is that issuing primary shares in an IPO may lead to greater regulatory scrutiny and a lengthy review process. To avoid Since 2005 the underwriters of IPO firms adopting either auction or bookbuilding have had the option of using pure auction/bookbuilding without reserving shares for the tranche of fixed-price public offer

11 administration procedures, issuing firms prefer selling secondary shares to investors. For the period , IPO firms in Taiwan just used either pure fixed-price public offer or hybrid auction (Chen and Wu, 2015). The use of secondary shares in bookbuilding was allowed in 2004 so that hybrid bookbuilding was effectively available for IPO firms since then. In 2005 the regulation was revised and stipulated that only secondary shares are used in IPOs. From then on, bookbuilding became the major underwriting procedure in Taiwan. In our sample period, there were pure fixed price public offer, pure bookbuilding offer, and hybrid bookbuilding. Auction was almost not used. In fixed price public offers, individual investors are potential subscribers who submit indication of buying shares up to three round lots (1,000 shares per round lot). Institutional investors do not show any interest in participating in fixed price public offers because of the constraint on order size. IPO lottery is used to determine share allocation in the case of oversubscription. The offer price is negotiated and set by both underwriters and IPO firms. In bookbuilding offers, underwriters set a preliminary offer price range in the first place. Investors are then invited to bid for the shares within the price range given. The offer price is set based on the demand for and supply of the IPO shares. To avoid ownership concentration, each investor is restricted from subscribing more than 3% of the shares offered. Although individual investors are not excluded from bookbuilding offers, the major participants are institutional investors. Before 2005 the hybrid bookbuilding was performed by soliciting indication of interests of investors at possible price levels first. The offer price was set after bookbuilding; the same offer price was used in the tranche of fixed price public offer. The investors in the tranche of public offer have the information of not only the preliminary offer price range but the specific offer price when they subscribe to the reserved shares. In 2005, to shorten the whole procedure, the hybrid bookbuilding was carried out by synchronizing the bookbuilding and public offer tranches. That is, the investors in the public offer tranche only know of the preliminary offer price range but not the specific offer price. 12 Figure 1 presents the detailed timeline. Moreover, to prevent the possibility that underwriters purposely lower the offer price to benefit bookbuilding investors, the rule in 2011 stipulated that both the lower bound of the preliminary offer price range and the offer price could not be less than 70% of the average of pre-ipo trading prices in the last ten days. For pure fixed price public offers, there is no specific regulation for setting the offer price to be related to pre-ipo 12 In the fixed price public offer tranche, investors without any knowledge of the specific offer price must deposit the dollar amount computed at the upper limit price in the preliminary offer price range

12 trading prices. 3. Literature Review and Hypothesis Development Pre-IPO or when-issued markets allow investors to trade for the shares of the firms that are about to go public. However, whether a pre-ipo market facilitates price discovery for IPO shares remains a puzzling issue. Unlike Taiwan where 3 (6)-month pre-ipo market trading has been mandatory for firms seeking IPOs since 2003, the U.K. gave firms an option to have their shares traded on the Alternative Stock Market before getting listed (Derrien and Kecskés, 2007). Derrien and Kecskés (2007) document that only 66 IPO firms chose to have an introduction on AIM before their listing. These two-stage IPO firms have the average initial return of 11.9%, which is significantly lower than 24.7% for the remaining 786 direct IPO firms. They argue that the two-stage offering reduces valuation uncertainty and enhances price discovery for the IPO firms, in turn, leads to a reduction of the initial returns. In Europe, the pre-ipo markets are over-the-counter and vary widely from country to country (Cornelli, Goldreich and Ljungqvist, 2006). The trading is typically organized by independent brokers and dominated by retail investors (Cornelli, Goldreich and Ljungqvist, 2006; Dorn, 2009). Moreover, the duration of pre-ipo market trading typically lasts for only a week or so before the first IPO trading day. Cornelli, Goldreich, and Ljungqvist (2006) indicate that investor sentiment affects the price discovery function of pre-ipo market: in high-sentiment market 13 the pre-ipo prices are not only higher than offer price but also the initial price (e.g. Dorn, 2009). Whether the high pre-ipo prices affect the offer price setting depends on the relative bargaining power between issuing firms and associated underwriters. The offer price tends to be set higher when issuing firms have higher bargaining power, and that results in lower initial return. By contrast, the offer price tend be set close to the middle point of the price range from bookbuilding procedure when underwriters have higher bargaining power, and that results in higher initial return. Moreover, when offer prices being set higher due to raising pre-ipo prices in high sentiment market, the over-boosted price would be followed by long-run reversal and therefore long-run underperformance. Dorn (2009) explores the German s when-issued market and find that 91 IPO firms choosing when-issued market trading 13 Cornelli, Goldreich, and Ljungqvist (2006) take the midpoint of the filing range as a proxy for the underwriter s ex ante prior of the fundamental value, and they define high sentiment is that pre-ipo price is higher than midpoint of the filing range

13 before IPO are associated with the average initial return of 50%, which is lower than the average initial return of 54% for the rest 36 IPO firms choosing direct IPO. Since the when-issued market is mainly comprised of individual investors, high investment sentiment would boost pre-ipo price as well as the initial price. The overshooting in prices would be followed by long-run underperformance. Chang et al. (2016) use 218 Taiwanese IPOs in the sample period and find that the pre-ipo market facilitates price discovery in the sense that the pre-ipo price is predictive to the initial price. However, the average initial return of 55% seems to be incongruent with the alleged merit associated with pre-ipo market. They refer the argument put forth by Ritter (2011) that information asymmetry only comprises a small portion of initial return. Rather, the agency problem manifested in underwriters rent-seeking behavior is the main driver of initial return. Moreover, Chang et al. (2016) find that both initial return and price error are significantly positive, implying that the pre-ipo price is not only higher than offer price but also initial price. 14 We briefly summarize the similarities and differences from the above studies regarding the pre-ipo market price, offer price and initial price. The similarities include (S1) the pre-ipo price is higher than both the offer price and initial aftermarket price, suggesting that overvaluation may exist in the pre-ipo market; (S2) the initial price would gradually converge to the pre-ipo price. The differences include (D1) the initial return would be lower (higher) when the offer price is (not) set in accordance with the boosted pre-ipo price; (D2) whether the exitance of pre-ipo market could facilitate price discovery for IPO shares remain a debatable issue. In this study we propose the self-dealing hypothesis to analyze the relationship among the pre-ipo market price, offer price, and initial aftermarket price. Moreover, we argue that the relation among these prices in bookbuilding offers is different from that in fixed price public offers. Those hypotheses are developed as follows. 3.1 Pre-IPO trading price and initial aftermarket price There are three parties involved in an IPO process: the issuer, investors, and the associated underwriter. The issuer would like to set a higher offer price for raising more proceeds from the IPO and to have a higher initial market price that reflects the wealth of shareholders. Based on the self-dealing motive, the issuer is likely to boost the pre-ipo price at a higher level so that both the offer price and initial aftermarket 14 Price error is the ratio of the pre-ipo price on ESM over the closing price on the first trading day on TWSE or GTSM. The statistics in Table 3 of Chang et al. (2016) indicate that the price errors on one day and four days prior to IPO are significantly positive

14 price could be higher than otherwise when the pre-ipo market price is used as a benchmark. The boost of pre-ipo prices in Taiwan could be attainable based on three facts. First, pre-ipo market trading is mainly comprised of individual investors who tend to be noisy traders (Chiang, Qian, and Sherman, 2010). Second, trading in the pre-ipo market is relatively inactive except for the first trading day. 15 Third, there is no price limit in pre-ipo trading in Taiwan s EMS. The three facts together make the EMS vulnerable to insider manipulation. In addition, the theory of information cascade suggests that investors who observe the investment choice made by previous investors can update their beliefs about the value of IPO shares (Welch, 1992). If pre-ipo market trading prices are boosted, these prices may become reference prices or signals to retail investors. Therefore, pre-ipo market prices tend to be positively related to aftermarket prices and lead to overreaction in the aftermarket. If offer prices are adjusted upwards under the circumstance of artificially boosting pre-marker prices, the short-run initial returns may be lower. Combining the issuer s self-dealing motive and the theory of information cascade, we develop the following testable hypothesis. Hypothesis 1: The pre-ipo price is higher than both the offer price and the initial aftermarket prices. The initial aftermarket prices would gradually converge to the pre-ipo market price. 3.2 Pre-IPO trading price and offer price From the perspective of an associated underwriter, setting the offer price is a complex task. The underwriter tends to provide a preliminary offer price range based on the issuing firm s fundamentals. Moreover, the pre-ipo market prices give the underwriter alternative information in setting the offer price. If underwriting fee that is in proportion to the offering proceeds is of concern, the underwriter would prefer setting a higher offer price. Alternatively, the underwriter may prefer setting a lower offer price to minimize underwriting risks. Whether the offer price is set with reference to the preliminary offer price range or the pre-ipo price depends on the issuer s attributes and the relative bargaining power between the underwriter and issuer. 15 According to Chang et al. (2016), the ESM trading in the first 6 months is inactive. The average daily dollar volume fluctuates around NT$ 4.1 million and the average daily turnover is around 0.07%, an annualized rate of about 17%. The median values are even lower

15 Both informed and uninformed investors would prefer a lower offer price for their best interests. However, because the pre-ipo market is associated with a short-sales constraint and a restriction on share subscription, the pre-ipo market in fixed price public offers is typically comprised of uninformed individual investors. For these uninformed individual investors, pre-ipo market prices provide the major information for them to judge whether the offer price and the initial aftermarket price are reasonable. Therefore, the offer price could be set with reference to the boosted pre-ipo market price. In a nutshell, for a fixed price public offer, setting the offer price at a higher level is for the best interests of the issuing firm and underwriter, and is also acceptable to uninformed individual investors. Because the offer price as well as the initial aftermarket price are set higher when the pre-ipo market price is used as a benchmark, we posit that the initial return calculated from the offer price to the initial aftermarket price would be lower. In a bookbuilding offer, the major participants are institutional investors. To deal with these informed institutional investors, the lead underwriter tends to set a lower offer price to benefit good-relation clients in exchange for side payments (Ljungqvist, 2007; Loughran and Ritter, 2002; Reuter, 2006; Ritter, 2011; Nimalendran, Ritter and Zhang, 2007; Liu and Ritter, 2010 and 2011; Goldstein, Irvine and Puckett 2011). The offer price is set in general with reference to the middle of offer price range while not in tandem with pre-ipo market prices. 16 Setting the offer price at a lower level would be associated with a higher initial return. Moreover, the relative bargaining power of underwriters also affects the setting of offer price. 17 In the IPO literature, Logue (1973) posits that relative bargaining power between issuers and underwriters influences the magnitude of underpricing. The underwriter s bargaining power depends on his/her contribution to the underwriting process and the client s dependence on the resources contributed by the underwriter. By contrast, the issuer s bargaining power depends on the value of the assets, rent generating capacity of the equity for sale, and financial independence (Marshall, 2004). Prior studies have shed light on the measurementsof relative bargaining power between issuers and underwriters Cornelli, Goldreich, and Ljungqvist (2006) indicate that the midpoint of the filing price range is a proxy of the issuer s fundamental value assumed by the associated underwriter. 17 Many threads of literature such as international joint ventures, strategic alliances, and cooperative agreements indicate that the relative bargaining power of the parties will have influences on the control over decision making and allocation of equity or proceeds (Coff, 1999; Harrigan & Newman, 1990; Inkpen & Beamish, 1997; Khanna, Gulati, & Nohria, 1998; Lecraw, 1984; Yan & Gray, 1994). 18 Logue (1973) proposes a dichotomous variable based on whether the underwriter is on a list of prestigious underwriters produced by Hayes (1971). For issuer bargaining power, he proposes that the issue size could serve as a proxy. Carter and Manaster (1990) use the proxy of the ranking position of an underwriter s name within tombstone ads (Hayes, 1971)

16 In general, the underwriters of bookbuilding offers have relatively larger bargaining power against the issuers than those of fixed price public offers do. This is because bookbuilding offers require higher placement skills and capability to maintain good relations with important clients than fixed price public offers do. 19 This larger bargaining power would motivate the underwriters to set lower and rational offer prices. There are two major pieces of information for them to set offer prices: the pre-ipo market prices and the indication of interests from bookbuilding. Because of investment sentiment and/or insider manipulation, the pre-ipo market prices tend to be boosted to a higher level than the indication of prices collected from bookbuilding. 20 We therefore posit that the offer price of a bookbuilding offer is set close to the prices collected from bookbuilding, while the offer price of a fixed price public offer is set close to the pre-ipo market prices. Furthermore, investment sentiment would prevail in both the pre-ipo market trading and the initial aftermarket trading. Because the offer prices of bookbuilding offers are set with reference to bookbuilding procedure and are much lower than the offer prices of fixed price public offers set with reference to pre-ipo market prices, the short-run initial returns of bookbuilding offers would be higher than those of fixed price public offers. Moreover, the initial aftermarket trading is also subject to investor sentiment which results in initial overreaction (Cornelli, Goldreich, and Ljungqvist, 2006). This initial overreaction will be corrected in the long run. We posit that the initial overreaction is higher for fixed price public offers than for bookbuilding offers. Therefore, the long-run stock performance of bookbuilding offers would be better than that of fixed price public offers if the initial overreaction is corrected in the long run. To sum up, we develop the following testable hypotheses. Hypothesis 2.1: The offer prices of fixed price public offers are set close to pre-ipo market prices, which are much higher than the offer prices of bookbuilding offers. Hypothesis 2.2: The initial returns of fixed price public offers are lower than those of bookbuilding offers. 19 Fagre and Wells (1982) indicate that bookbuilding underwriters can access to public equity markets through their sales force and social and business networks. Moreover, their local knowledge of large institutional investors allows them to precisely apprehend demand through the bookbuilding process (Hoberg, 2007). Furthermore, the skill of using market knowledge during pricing negotiations can be a resource that adds to bargaining power and advantage (Coff, 1999; Inkpen & Beamish, 1997; Lecraw, 1984; Schelling, 1956). 20 Using pre-ipo market prices as a proxy of small investors valuation or sentiment, Cornelli, Goldreich, and Ljungqvist (2006) find that high pre-ipo market prices are a very good predictor of first-day aftermarket prices, while low grey market prices are not. Moreover, long-run price reversal only follows high grey market prices

17 Hypothesis 2.3: The long-run stock performance of fixed price public offers are worse than that of bookbuilding offers. 4. Data, Variables, and Models Our sample consists of 476 IPO firms in the period of The data for ESM, TWSE, and GTSM including prices, trading volume, and the number of shares outstanding are collected from the Taiwan Economic Journal (TEJ), a data vendor in Taiwan. Moreover, other data items such as firm age, assets, directors shareholdings, duration, leverage, and EPS are also collected from TEJ. IPO characteristics such as offer prices, the number of shares issued, underwriters, and auditing firms are collected from the IPO prospectus. All sample firms have been subject to a regulation of ESM trading for at least 3 months since 2003 and 6 months since Table 1 reports the annual sample distribution and the average initial returns for fixed price public offers and bookbuilding offers, respectively. The results show that the number of firms choosing the fixed price public offer is significantly reduced from 80 in 2003 to 1 in Starting from 2007, there was no IPO firm choosing the fixed price public offer. In contrast, bookbuilding has been graduately accepted by IPO firms as the only underwriting procedure. We note that most bookbuilding IPOs in our sample are hybrid bookbuilding offerings with two tranches. 21 The process of bookbuilding tranche usually lasts for four business days, and the process of fixed price offering tranche often starts one day later than the bookbuilding tranche but ends at the same time. The announcement of bookbuilding is accompanied with a suggested price range. The final offer price is determined one day after the end of the bookbuilding process. We find that the average initial return of fixed price public offers (8.9%) is much lower than that of bookbuilding offers (44.9%). <<Insert Table 1 Here>> 21 Among the 322 IPOs associated with bookbuilding, 313 IPOs are hybrid bookbuilding offerings and 9 IPOs are pure bookbuilding ones

18 We provide the description of each variable in Appendix A. Table 2 reports the summary statistics for variables. The initial return which is the percentage change from the offer price to the initial closing price is 33.3% on average. The average market-adjusted initial return is 42.0%. We also use the price-multiple approach in Purnanandam and Swaminathan (2004) to estimate the fundamental offer price that are derived from the price multiples of matching firms which are listing firms, close in asset size, and in the same industry as the IPO firm. Specifically, the fundamental offer price (OPR) is obtained from the following condition. (1) We then calculate the underpricing of the offer price (Undermatch) with respect to the fundamental offer price as. (2) The average underpricing of offer price with respect to the fundamental offer price is -16.9%, suggesting that the original offer price is not underpriced on average. That is, the typical original offer price is set much higher than the estimated fundamental offer price. The setting of offer price at a relatively low level would result in a price increase when shares are initially traded in the aftermarket (Aggarwal and Rivoli (1990), Ljungqqist et al. (2006), Baker and Wurgler (2006), Dorn (2009) and Clarke et al. (2016)). Therefore, we also calculate the overreaction of the initial price with respect to the matching firm (Overmatch). Again, we use the price multiple approach in Purnanandam and Swaminathan (2004) and estimate the fundamental initial price with respect to the matching firm. The overreaction of the initial price is calculated as. (3) The average overreaction of the initial price with respect to the matching firm is 50.6%. The results from the overestimation of offer price and the overreaction of the initial price suggest that both the pre-ipo market and the post-issuance market in Taiwan may be affected by high investor sentiment that boosts the level of offer price as well as the initial closing price