Friends Can Help: The Effects of Relationships in the Chinese. Book-Building Process

Transcription

1 Friends Can Help: The Effects of Relationships in the Chinese Book-Building Process Wei Luo Heng Yue* Peking University Lu Zhang Beijing Industry and Commerce University January 16, 2015 Abstract In this paper, we investigate the extent to which the business relationship with an underwriter affects an institutional investor s bidding behavior in the book-building process. Using detailed bidding information in China, we find that when institutional investors have a close business relationship with the underwriter, they are more likely to participate in bidding and their bidding prices are more optimistic than those of other institutional investors. The underwriter makes use of the optimistic bidding price to set high issuing prices for IPO firms. Moreover, these effects are more pronounced for IPO firms with inferior earnings quality. We also document that the adjustment of the issuing price is negatively related to post-ipo stock returns. The evidence indicates that the relationship between institutional investors and the underwriter reduces the efficiency of the IPO pricing process. JEL classification: G15 G24 G30 M41 Keyword: Book-building; Relationship; Firm Quality; IPOs in China * Corresponding author Acknowledgement: The authors would like to thank Hai Lu, Qingquan Xin, and seminar participants at Chongqing University, Peking University, Taiwan University, and Zhongnan University of Economics and Law, for their insightful comments. The authors appreciate financial support from the National Natural Science Foundation of China (Approval No and ). 1

2 Friends Can Help: The Effects of Relationships in the Chinese Book-Building Process Abstract In this paper, we investigate the extent to which the business relationship with an underwriter affects an institutional investor s bidding behavior in the book-building process. Using detailed bidding information in China, we find that when institutional investors have a close business relationship with the underwriter, they are more likely to participate in bidding and their bidding prices are more optimistic than those of other institutional investors. The underwriter makes use of the optimistic bidding price to set high issuing prices for IPO firms. Moreover, these effects are more pronounced for IPO firms with inferior earnings quality. We also document that the adjustment of the issuing price is negatively related to post-ipo stock returns. The evidence indicates that the relationship between institutional investors and the underwriter reduces the efficiency of the IPO pricing process. JEL classification: G15 G24 G30 M41 Keyword: Book-building; Relationship; Firm Quality; IPOs in China 2

3 Friends Can Help: The Effects of Relationships in the Chinese Book-Building Process 1. Introduction When a firm publicly issues shares for the first time (i.e., an IPO), there is great uncertainty regarding the market value of the firm. To achieve an appropriate issuing price, a variety of IPO mechanisms have been designed, among which book-building is the most widely used. In the standard book-building process, underwriters invite prospective institutional investors to submit their bidding prices and demanded quantities for the IPO shares. The invited institutional investors then collect and provide information about the value of the IPO firm. The underwriters decide the IPO price and allocate IPO shares at their discretion to the institutional investors. Previous studies offer different views on the share allocation discretion enjoyed by the underwriters. Some studies (e.g. Benveniste and Spindt, 1989; Sherman and Titman, 2002; Ljungqvist and Wilhelm, 2002; Cornelli and Goldreich, 2003; Bertoni and Giudici, 2014) suggest that the allocation discretion of the underwriters significantly improves the efficiency of IPO prices by inducing optimal information production. More recent studies show that the underwriters can opportunistically use their discretion on IPO share allocations in exchange for future private benefits from related institutional investors (Loughran and Ritter, 2002; Reuter, 2006; Jenkinson and Jones, 2009; Liu and Ritter, 2010; Goldstein et al., 2011). This opportunistic behavior is promoted by the combination of underwriters allocation discretion of 1

4 shares, and the relationship between the underwriters and the institutional investors. However, it is unclear whether and how the relationship between the underwriters and institutional investors shapes the book-building process when the underwriters have only the price discretion but no allocation discretion, as the case in China. In this paper, we examine whether institutional investors who have a good relationship with the underwriters intentionally provide high bidding prices to help the underwriters boost the issuing price. Our research question is motivated by a large amount of anecdotal evidence in China. For example, at The tenth international forum of investment funds on Dec. 2, 2011, Yujun Zhang, the CEO of the Shanghai Stock Exchange, and Liping Song, the CEO of the Shenzhen Stock Exchange, both criticized the over-bidding of institutional investors in the IPO process and the existence of relationship effects in bidding. 1 In early 2012, a survey of 67 Chinese mutual fund companies confirmed this criticism. Approximately 48 percent of mutual fund managers admitted that they would bid high prices in exchange for services from a brokerage firm to maintain a personal network or to appease the fund shareholders who have a good relationship with the brokerage firm. 2 A press article also reported that some institutional investors received calls from the underwriter requesting them to raise their bidding prices. 3 In this paper, we use a large sample of archival data to directly examine the bidding behavior of institutional investors and the effect that their relationship with the underwriter has on the book-building process. 1 Mainland China has only two stock exchanges: the Shanghai Stock Exchange and the Shenzhen Stock Exchange. Reports on this forum (in Chinese) can be found at 2 The survey was conducted by Security Times, a semi-official newspaper that is authorized by the regulator to publish annual reports and other corporate disclosures. The article link (in Chinese) is

5 The underwriters have a strong incentive to set a high issuing price due to the fee structure of underwriting. In China, when the issuing price is higher than the reservation price of the issuing firm, the underwriters charge additional incentive fees. The incentive fees can reach as high as 13 percent of the total IPO proceedings. However, the price discretion of the underwriters is constrained by the institutional investors who participate in the book-building and provide the bidding prices. If the institutional investors provide low bidding prices, the Chinese regulations stipulate that the underwriters can only set a low price. In contrast, if the institutional investors provide high bidding prices, the underwriters have an excuse to set a high offer price. Therefore, the institutional investors can help the underwriters achieve high offer prices (and high underwriting fees) by providing high bidding prices. The related institutional investors also have incentives to help the underwriters even though the underwriters have no allocation discretion in the book-building process. First, the underwriters are usually brokerage firms that provide services to institutional investors. In addition to needing favorable analyst reports from brokerage firms (Gu et al., 2013), the institutional investors also rely on the channels of the brokerage firms to market their new funds to retail investors. Therefore, the institutional investors are willing to help the underwriters in order to maintain their good relationship. Second, the managers of mutual funds often have a personal relationship with the managers of brokerage firms. Many fund managers are former analysts from brokerage firms. This social tie is beneficial when the underwriters need support from the institutional investors. Third, the shareholders of fund management 3

6 companies are closely related to the shareholders of the brokerage firms in China. Over 60 percent of the fund management companies in China have brokerage firms or their parent companies as their shareholders. Providing a high bid price can be costly for institutional investors if they have to buy the stocks at the inflated prices. However, the allocation mechanism in China significantly mitigates the concerns of related institutional investors. Because Chinese IPOs are always over-subscribed, new shares are allocated by lottery. For instance, 86 institutional investors participated in the book building of Fuling Zhacai Company (Stock code: ). The highest bid was 35.7 percent higher than the final price of the IPO. After the lottery process, the institutional investor that provided the highest bid did not receive any shares. Thus, the over-bidding institutional investors may not actually buy the overpriced stocks. In addition, even if these institutional investors win the lottery to buy the IPO shares, they pay at the offer price rather than their bidding prices. Therefore, the probability of actually paying a high price for the IPO shares is low if the related institutional investors try to help the underwriter. To empirically test whether such opportunistic behavior exists, we analyze a unique database that includes the detailed bidding information of Chinese institutional investors in the IPO book-building processes during the period of November 2010 to April We use two proxies of the relationship between an underwriter and an institutional investor: (1) the trade commission paid by the institutional investor to the underwriter, and (2) the number of new funds issued by the institutional investor around the time of the IPO book-building. A high trade commission suggests a good 4

7 relationship between the institutional investor and the underwriter. The need for marketing channels for new funds gives the institutional investor an incentive to cater to the underwriter. We find that after controlling for other potential determinants of the bidding characteristics, institutional investors in a close business relationship with an underwriter are more likely to participate in the book-building process. Moreover, these related institutional investors provide more optimistic bidding prices than other institutional investors. Further tests show that the underwriters take advantage of related institutional investors optimistic bids to increase the issuing prices. The evidence is consistent with the hypothesis that related institutional investors help the underwriters to boost the issuing price. We further investigate the effects of issuers earnings quality. We find that the relationship effects mainly exist when the IPO firms have low earnings quality, as in this case the underwriters have greater difficulty in achieving high issuing prices. The evidence suggests that related institutional investors provide the greatest support when the underwriters need them the most. An alternative explanation for the above findings is that related institutional investors bid higher because they gain private information from the underwriters, especially when the issuers have lower earnings quality. We thus examine the effect of upward adjustments in the offer price. Our analyses show that the degree of upward adjustment in the offer price has no significant effect on IPO underpricing, but is negatively associated with post-ipo stock returns. The results indicate that other investors suffer from the high bids of related institutional investors, suggesting that 5

8 the relationship between the institutional investors and the underwriters reduces the pricing efficiency in the IPO process. Our paper contributes to the literature in the following ways. First, our findings extend our understanding of the business relationships in the book-building process. Previous studies mainly focus on the role of business relationships in the allocation of IPO shares (e.g., Reuter, 2006; Jenkinson and Jones, 2009; Goldstein et al., 2011). We use detailed hand-collected bidding information and provide direct evidence on the role of business relationships in deciding the offer price. Our evidence suggests that an institutional investor s relationship with an underwriter is an important factor in determining the investor s bidding activities in China. Second, we propose and test a form of opportunistic behavior in the book-building process when the underwriters have price discretion but no allocation discretion. The book-building literature regards information revelation as the benefits of pricing discretion and/or allocation discretion (e.g., Benveniste and Spindt, 1989; Benveniste and Wilhelm, 1990; Sherman and Titman, 2002; Sherman, 2005). We find that related institutional investors provide misleading information in their bids, especially for IPO firms with low quality earnings. The opportunistic behavior casts doubt on the efficiency of the information production during the book-building process. Third, our findings have implications for the earnings quality literature on IPO firms. Prior literature shows that IPO firms have incentives to manage earnings, which leads to low earnings quality before the IPO (see Dechow et al., 2010). We provide 6

9 evidence on how institutional investors react to the earnings quality of issuing firms during the book-building process. Our results indicate that the business relationship is more detrimental to the pricing efficiency of the IPO process when the issuing firm has low earnings quality. Fourth, our findings have important policy implications for China and other countries. Although Chinese regulators have tried to restrict the possible opportunistic allocation of IPO shares by the underwriters, our evidence suggests that a different type of opportunistic behavior still exists, calling for further consideration of the design of the IPO process. Our results indicate that the collusion between the underwriters and related institutional investors undermines the efforts of the regulator. The rest of this paper is organized as follows. In Section 2, we provide an overview of China s book-building system and develop the hypotheses. Our data sources, measurements of key variables and descriptive statistics are presented in Section 3. Our main regression results are discussed in Section 4. In Section 5, we explore the effects of optimistic bids on IPO pricing and firms post-ipo performance. Section 6 concludes. 2. Institutional Background and Hypothesis Development 2.1 Institutional Background: The book-building process in China The Chinese stock market was established in the early 1990s. In the relatively short period since then, the IPO issuance mechanisms have undergone many changes. Initially, the regulator used fixed price-earnings ratios to set the IPO issuing prices. 7

10 Before 1998, the IPO pricing was set at a fixed P/E ratio of 15. From 1998 to 2005, the regulator allowed variations in the P/E ratios, but set 20 as the upper limit. The fixed P/E ratio method cannot effectively determine the real value of IPO firms. To improve the market-oriented pricing mechanism, the China Security Regulatory Commission (CSRC hereafter) launched a series of reforms regarding the IPO process and formally introduced the book-building mechanism in January Book-building refers to the process of generating, capturing, and recording the demand for new shares from institutional investors to support efficient price discovery. However, the book-building system in China has a number of special features. First, every public offering in China consists of offline rationing and online issuance. The offline rationing is only open to institutional investors who take part in the book-building process. The online issuance is open to individual investors and other institutional investors. This separation is similar to the system adopted by the Hong Kong stock markets. The shares available to offline institutional investors in Hong Kong are approximately 90 percent of the total new shares, however, while in mainland China, the ratio is no more than 50 percent and can be as low as 20 percent. 4 The limited shares offered offline affect the incentives of institutional investors in the book-building process, because they may receive very few shares, in which case the benefits would not cover their costs of research on the IPO firms. Therefore, institutional investors usually put few efforts to investigate the real value of IPO firms. 4 On April 28, 2012, the CSRC changed its policy on institutional investors and required the offline rationing to be greater than 50 percent of total new shares. The latest regulation issued on December 13, 2013, further increased the percentage to greater than 60 percent for small issuers and greater than 70 percent for large issuers. 8

11 Second, the underwriters cannot determine who partakes in the bidding. Underwriters can recommend institutional investors to participate in the book-building process. But they cannot decline the participation of any qualified institutional investor. Qualified institutional investors can voluntarily participate in the book-building process and provide bidding information with prices and the demanded quantity. Furthermore, the Chinese book-building system does not give the underwriter the right to allocate IPO shares in the offline rationing. If the new shares are oversubscribed, the shares will be allocated by lottery to institutional investors who provide bidding prices higher than the issuing price. In China, nearly all stocks are oversubscribed, 5 and some stocks are oversubscribed by more than 10 times. Due to the limited availability of new shares and the lottery allocation process, institutional investors have less incentive to investigate the real value of IPO firms. Thus, many institutional investors do not provide informative bidding prices. The variance of the bidding price demonstrates the seriousness of this problem. As Li et al. (2012) document, the dispersion in bids is extremely high in China. For instance, forty-two investors participated in the book-building of Guodian Qingxin Company (Stock code: ). The maximum bid price was RMB67.5, four times the minimum bid price of RMB17.6. Because some institutional investors do not conduct their own research, they may be misled by others who have an incentive to report high bidding prices. 5 In the history of China s stock markets, Nanning Baling is the only cancelled IPO that failed to attract sufficient bids from investors. 9

12 To summarize, the regulatory rules in China put constraints on underwriters discretion and reduce institutional investors incentives to provide valuable information in the book-building process Hypothesis development In the standard book-building process, underwriters take advantage of allocation rights to facilitate information generation during the process of going public. A number of book-building models have been developed in theoretical studies, such as Benveniste and Spindt (1989) and Benveniste and Wilhelm (1990), which note that informed investors are given priority in the distribution of shares to reveal what they know to the underwriter. Sherman and Titman (2002) extend the theory and demonstrate that allocation discretion substantially decreases information asymmetry and improves pricing accuracy. Cornelli and Goldreich (2001, 2003) and Ljungqvist and Wihelm (2002) provide empirical support for these models. The relationship between underwriters and institutional investors, however, also raises concerns in regard to conflicts of interest. Loughran and Ritter (2004) examine the agency problems between the underwriter and the issuer. For their private benefit, underwriters allocate more hot and underpriced IPOs to executives of the issuer and/or institutional investors with whom they have a good relationship. Liu and Ritter (2010) find that the issuers whose top executives receive a hot IPO allocation are less likely to switch underwriters in a follow-on offer. In addition, underwriters may sacrifice the issuer s interests for the trading commissions from institutional investors. 10

13 Reuter (2006) documents that the more commissions investors pay to the lead underwriters, the more shares of a hot IPO they receive. Jenkinson and Jones (2009) survey approximately four hundred institutional investors and report that the brokering relationship with the underwriter is perceived to be the most important factor influencing share allocations. Nimalendran et al. (2007) also find that investors trade liquid stocks to influence their IPO allocations. The agency perspective indicates that giving discretionary allocation power to the underwriter creates conflicts of interest and induces the underwriter to act opportunistically. It is unknown whether removing allocation discretion would reduce the possible opportunistic activities stemming from the close relationship between the underwriters and institutional investors. In China, the regulator does not give allocation discretion power to the underwriter to reduce the possible opportunistic activities of the underwriter. However, we argue that another form of opportunistic activity still exists. That is, the underwriter may ask related institutional investors to boost the issuing price. In China, the underwriting fee consists of two parts. An underwriter first charges a fixed underwriting fee for the IPO. The underwriter will also charge a variable fee based on a percentage of the total proceedings from the IPO. The higher the issuing price, the more proceeds the issuing firm raises and the more fees the underwriter earns. More importantly, the variable fee is aggressive and can be as high as 13 percent of the proceeds. Therefore, the underwriting fee structure creates a strong incentive for the underwriter to boost the issuing price. 11

14 However, the regulations in China stipulate that the underwriters cannot decide the price entirely at their will. The underwriters must consider the bidding prices of the institutional investors who participate in the book-building process. 6 Therefore, the underwriter needs help from institutional investors to reach a high issuing price. We argue that the underwriter can gain help from institutional investors to boost the issuing price. First, the institutional investors and the underwriter who is also the brokerage firm, often have a close business relationship. For example, the institutional investors (mostly mutual funds) use the underwriter (also the brokerage firm) as the dominant channel to market their new funds. The institutional investors also rely on the analysts in brokerage firms for timely and accurate research information on their holding stocks. More importantly, the institutional investors need favorable recommendations and reports on stocks heavily held in their portfolios to maintain or boost the value of those stocks (Firth et al., 2013; Gu et al., 2013). Therefore, when the underwriter needs help in boosting the issuing price, the institutional investors may return the favor by providing more optimistic bids. Second, although overly optimistic bidding may be costly to institutional investors, the cost is not large and may be irrelevant to fund managers. Because the new shares are allocated to institutional investors by lottery and the final buying price is the issuing price not the bidding price, the possible loss from the overpriced shares is relatively small. Moreover, as aforementioned, the majority of new shares are 6 See, for example, Administration Measures on Securities Issuance and Underwriting (CSRC Decree No.37) and Decision on Amending the Measures for the Administration of Securities Offering and Underwriting (CSRC Decree No 69). 12

15 allocated to individual investors. The institutional investors need to buy as little as 20 percent of the new shares, which also reduces the possible loss the institutional investors need to bear. In addition, the fund managers may not be overly concerned about the fund performance, because their compensation is based on the fund size, not the fund performance. Third, business in China is primarily relationship-based, unlike the rules-based business environments in Western countries (Liu et al., 2012). Accordingly, the underwriter may offer benefits to the fund managers in hidden ways. The bidding strategy is at the sole discretion of the fund management companies. Without legal constraints, the managers of the institutional investment firms are prone to conspire with the underwriters. We therefore expect that institutional investors who have a good relationship with the underwriter are more likely to participate in the bidding process and submit relatively optimistic bidding prices. Our first hypothesis is as follows: H1: Ceteris paribus, institutional investors who have a relationship with the underwriter are more likely to participate in the bidding process and provide an optimistic bidding price. The pricing of IPOs is a particularly challenging process because, prior to going public, investors have little information about the issuers. The information asymmetry during an IPO is severe. Thus, investors must rely heavily on the information in the IPO prospectuses, which typically includes only three years of annual and quarterly 13

16 financial statements, to value the IPO issuers and make their investment decisions. The importance of earnings in IPOs provides incentives for issuing firms to upwardly manage their earnings. Numerous studies have proved that issuers use positive accruals to report favorable pre-ipo earnings. Teoh et al. (1998) document high discretionary accruals around the time of IPOs. DuCharme et al. (2004) note that abnormal accounting accruals are unusually high around the time of stock offerings and are especially high for firms whose offerings subsequently attract lawsuits. In addition, Darrough and Rangan (2005) document a positive association between discretionary current accruals in the offering year and managerial selling, providing evidence of managers opportunistic motivation to manipulate earnings during the IPO process. In China, there is a strong incentive to manage earnings, because the regulator has imposed a set of minimum profitability requirements for IPOs. For instance, to apply for an IPO on the main-board and the small and medium enterprises board (SME board) in China, a company must satisfy all of the following requirements: (1) it must report a positive net income for each of the most recent three years, and the cumulated amount of net income must be greater than RMB30 million for these years; (2) the cumulated net operating cash flow must be more than RMB50 million for the most recent three years, or the cumulated amount of revenue must be more than RMB300 million for these years; (3) the amount of common stock capital before the issuance must be more than RMB30 million; and (4) intangible assets, excluding land-use rights, the rights to water aquaculture and mining rights, must not be more 14

17 than 20 percent of net assets in the most recent year. Some firms that cannot meet the requirements operationally have to resort to earnings management (Chen and Yuan, 2004; Kao et al., 2009; Aharony et al., 2010). However, earnings management tends to reverse in the future and lead to lower future performance. Ragan (1998) and Teoh et al. (1998) find that issuers with aggressive earnings management may experience a sharp decline in long-term stock returns. As a result, institutional investors are concerned that artificially raised earnings will eventually reverse and, in turn, harm the performance of their portfolio. They are therefore unwilling to participate in book-building for IPO issuers with overstated earnings (Neupane et al., 2014). If they do participate, they submit bids with lower prices. Therefore, when IPO firms have low earnings quality, the underwriter relies more on the assistance of related institutional investors to successfully issue the stocks and/or achieve high issuing prices. Our second hypothesis is stated as follows: H2: The effect of the relationship on institutional investors bidding behavior, as hypothesized in H1, is more evident when the issuing firm has low earnings quality than when the issuing firm has high earnings quality. 3. Sample and Research Design 3.1 Sample and Data Source Our data period starts in November 2010 because the CSRC issued a new regulation at that time, which required IPO issuers and underwriters to disclose 15

18 relevant information about institutional investors bids during the book-building process. We hand collect institutional investors bid data 7 from the filings of Results on Offline Rationing by IPO firms. 8 We restrict our sample to IPO firms on the SME board and the New Growth Enterprise Board (ChiNext) during the period of November 2010 to April After deleting observations with missing values, we obtain a final sample of 139 IPOs on the SME board and 152 IPOs on ChiNext. Pre-IPO financial data and trading data in the secondary market come from the China Stock Market and Accounting Research (CSMAR) database. The WIND database provides the remaining data, such as brokerage fees and the size of the fund management firm. 3.2 Research Design We first examine whether the relationship with an underwriter affects the likelihood of an institutional investor participating in the book-building of an IPO. We use BID to measure whether the investor participates in the book-building. BID is a dummy variable that takes a value of 1 if an institutional investor submits a bid for an IPO and 0 otherwise. To measure the degree of optimism in the bidding price of each institutional investor, we need a benchmark to compare with. Our benchmarks are from the 7 These data are published by underwriters in filings named Announcement on Offline Lot and Allotment Result. 8 There are two stages in the book-building process. For IPO issuers on the main board, the first stage only provides a price range. After re-bidding, IPO issuers and underwriters determine an issuing price in the second stage. The SME board and ChiNext have different rules on book-building. For issuers on these two boards, the offering price is directly determined in the first stage. 9 The CSRC modified the rules on the offline rationing and lockup requirements in May In November 2012, the CSRC halted the IPO approval process. Our results remain the same after including 30 IPOs between May 2012 to November

19 valuation reports of the underwriters. According to CSRC regulations, the analysts of the underwriters need to provide a valuation report during the road show. The valuation report contains fundamental information about the firm and an initial valuation range. Our first benchmark is the fair value per share (FVE) estimated using the comparable valuation method and the information from the underwriter. Specifically, the fair value per share is estimated as follows: FVE = (1 Shareissuable% ) EPS P / E (1) where Shareissuable% refers to new shares issued, divided by shares outstanding after the IPO; EPS represents the basic EPS in the latest fiscal year disclosed in the prospectus; and P/E is the industry comparable P/E ratio as disclosed by the underwriter. Our first estimation of the optimism of the bid (Optimism1) is then calculated as the institutional investors bidding price deflated by the fair value per share (FVE). Another benchmark is the midpoint of the initial price range in the valuation report provided by the underwriter. Our second measure of optimism, Optimism2, is calculated as the institutional investors bidding price deflated by the midpoint of the initial price range in the valuation report of the underwriter. We use demand quantity as a weight to calculate the weighted average bidding price if there are several bids from one institutional investor. 10 We use two measures to proxy for the relationship between the institutional investor and the underwriter. The variable Relation is based on the trading 10 Some fund management companies submit several different bidding prices for an IPO firm. For instance, ICBCCS submitted three different bids for the Dahuanong IPO (CODE: ) underwritten by China Merchants Security. 17

20 commission information following Reuter (2006). Under the requirements of the China Security Law, fund management companies have to rent transaction seats from brokerage companies on the stock exchanges to buy or sell stocks. These rents are well-known as trading commissions. The choice of brokers and the trading commissions paid by fund management companies are influenced by the relationship between the institutional investor and the underwriter. If a fund management company has a good relationship with an underwriter, the company will give the underwriter more commissions than other fund managers. The Relation variable is constructed as follows. For each underwriter, we calculate the percentage of the trade commission paid by each fund management company over its total trading commission revenue. If a fund management company does not pay any trading commission to an underwriter, the percentage is set to zero. We then adjust the percentage of the trade commission by the median percentage for all fund management companies. Intuitively, our variable Relation measures whether a fund management company allocates above average commission fees to a brokerage firm. The higher the percentage, the closer the relationship between the institutional investor and the underwriter. Our alternative measure of the institutional investor-underwriter relationship is Fund_new, the natural logarithm of one plus the number of new funds distributed by an underwriter for a fund management company within two months before and two months after the month of the book-building of an IPO. Fund management companies often rely on the underwriter to market their new funds. If they intend to issue new funds, they are motivated to maintain a good relationship with the underwriter. 18

21 To test whether related institutional investors are more likely to participate in IPO book-building and submit more optimistic bids, we estimate the following Logistic model (2) and Tobit model (3), respectively. BID ααα_αααα_α Relationship DAC D Fbig UWR MKT IPO bf Size αααα ROE Lev Growth List Year Industry (2) Optimism Relationship DAC _ D Fbig UWR MKT IPO _ bf Over _ buy CV Size ROE Lev Growth List Year Industry (3) where α 1 captures the effect of the relationship (Relation and Fund_new) on the likelihood of the institutional investor participating in the book-building; and β 1 captures the effect of the relationship on the degree of optimism in bidding prices. Our first hypothesis predicts that both α 1 and β 1 will be positive. In line with prior studies, we use discretionary accruals to measure earnings quality. Following Kothari et al. (2005), we use the performance-matched modified-jones model: TAC 1 PPE Sale AR Asset t t t t Asset t 1 t 1 Assett 1 Asset t 1 (4) where PPE is property, plant, and equipment, ΔSale is the change in sales, and ΔAR is the change in accounts receivable. All of the variables are deflated by the beginning balance of the firm s total assets. Equation (4) is estimated by first excluding all IPO firms in year t, and the regression is then estimated by industry (two-digit CSRC industry code for the manufacturing industry and one-digit code for the remaining industries) and year. Any industry with fewer than five observations is removed. The residual term is the discretionary accrual for each individual firm. Finally, we match 19

22 each IPO firm observation with another firm from the same industry and year, and with the closest current year ROA. The difference between the discretionary accrual of the IPO firm and the matching firm is our measure of earnings quality (DAC). The higher the DAC, the lower the earnings quality. We then construct a dummy variable DAC_D, which equals 1 if the DAC is greater than the median of the sample and 0 otherwise. We winsorize the top and bottom 1% of observations for all of the continuous variables. To test whether the effect of the relationship on institutional investors bidding behavior is more evident when the issuing firm has low earnings quality than when the issuing firm has high earnings quality, we estimate the following Logistic model (5) and Tobit model (6), respectively. As hypothesized in H2, we predict that both α 2 and β 2 will be positive. BID ααα* Rα_αα elationship _ Relationship DAC D DAC D Fbig UWR αα_ααααα MKT IPO bf Size ROE Lev Growth List Year Industry (5) Optimism Relationship Relationship* DAC _ D DAC _ D Fbig UWR MKT IPO _ bf Over _ buy CV Size ROE Lev Growth List Year Industry (6) Following prior research (Liu et al., 2011; Yu et al., 2013), we also include a set of control variables and the fixed effect of the IPO firm s industry. Fbig measures the size of a fund management company, and is a dummy variable that equals 1 if the company is among the top 10 in terms of assets under management at the end of the quarter prior to the book-building and 0 otherwise. UWR measures the underwriter s reputation, and equals 1 if an underwriter is among the top 10 investment banks 20

23 measured by market share in the year prior to the IPO and 0 otherwise. We construct two measures of market conditions before the book-building of an IPO. MKT is calculated as the sum of the 10-day value-weighted market returns before the book building deadline. IPO_bf reflects the 10-day IPO activities before the book building deadline, and is the logarithm of one plus the number of IPOs. In addition, we directly measure investor sentiment in the IPO process as the offline over-subscription rate (Over_buy), which is the logarithm of the number of subscriptions divided by the number of shares offered. CV is the standard deviation of all of the investors bids, standardized by the mean of the bids. We also control the characteristics of IPO firms, including firm size (Size, logarithm of total assets at the beginning of the IPO year), profitability (ROE, return on equity in the year prior to the IPO), leverage (Lev, the ratio of total liability over total assets in the year prior to the IPO), growth (Growth, the rate of revenue growth), and stock exchange (List, equal to 1 if an IPO is listed on the SME and 0 otherwise). In addition to the analyses at the institutional investor level (i.e. each investor-bid is treated as one observation), we examine the effect of the institutional investor-underwriter relationship on the aggregated bidding behavior at the IPO firm level (i.e. each IPO firm is treated as one observation). Specifically, we run the following regression: Optimism _ c Relation _ c DAC _ D UWR MKT Overbuy CV IPO _ bf Size ROE Lev Growth List Year Industry (7) Optimism_c is the optimism at the firm level, which is calculated as the 21

24 quantity-weighted average of all bidding prices divided by the middle point of the initial price range. The higher the Optimism_c, the higher the bidding prices of the institutional investors compared to the underwriter s initial price. Our measures of the institutional investor-underwriter relationship at the IPO firm level are Relation_c1 and Relation_c2. Relation_c1 is measured as the total amount of trading commissions from all of the fund management companies that submit bids, scaled by the underwriter s total trading commission revenue. Relation_c2 is measured as the natural logarithm of one plus the total number of funds the underwriter distributes for all fund management companies submitting bids within two months before and two months after the month of the book-building of an IPO. These two variables measure the degree of the relationship for all participating institutional investors as a whole. This regression examines whether a closer relationship between the institutional investor and the underwriter leads to more optimistic bidding prices at the IPO firm level. 3.3 Descriptive Statistics Table 1 presents the descriptive statistics of our variables. Panel A shows the results at the institutional investor level. We observe that the mean value of BID is 0.269, indicating that institutional investors participate in 26.9 percent of the book-buildings. Optimism1 has a mean of 0.233, suggesting that the average bid of the institutional investors is 23.3 percent of the fair value of an IPO. This number is 22

25 small because we set the bidding price as 0 if an institutional investor does not participate in the book-build of an IPO. When we only consider the institutional investors that submit bid prices, the mean rises to The bid price can go as high as times the fair value. Optimism2 exhibits the same pattern. This suggests that investors have a large degree of uncertainty about the real value of IPOs (Li et al., 2012). The average value of Relation is 1.1 percent. We use this abnormal level of trading commissions to indicate whether there is a relationship between institutional investors and underwriters. The mean of Fund_new (0.647) suggests that each institutional investor consigns approximately 0.91 new funds to the underwriter within two months before and two months after the book-building of an IPO. [INSERT TABLE 1 ABOUT HERE] Of the total IPO sample, 40.5 percent were underwritten by a reputable underwriter. The mean value of MKT is negative, indicating that during our sample period, the stock market did not perform well. However, the IPO market was still hot. The mean of IPO_bf indicates that there are on average IPOs within 10 days of the book-building deadline. In addition, on average, an IPO is over-subscribed by times at the offline stage. The pre-ipo mean (median) ROE is (0.273), suggesting that the operating performance of IPO firms is superior. The mean value of total assets (Size) for issuers at the beginning of the IPO year is RMB million. Growth has a mean of 35.5 percent. The average of Lev is On average, IPO firms have positive pre-ipo discretionary accruals (DAC), indicating that the issuers 23

26 manage earnings upward prior to the IPO. In Panel B, we also measure optimism and the institutional investor-underwriter relationship at the IPO firm level. Optimism_c has a mean of 0.821, indicating that the bidding prices from all institutional investors (Optimism_c) are 82.1 percent of the middle point of the initial price range in the prospectus. Over_price (Over_price1 and Over_price2) is defined as the final issuing price divided by the average bids of all non-related institutional investors. Its mean of suggests that underwriters set the offer price 8.3 percent higher than the average bidding prices of all non-related investors. Relation_c1 and Relation_c2 describe the relationship between the underwriter and institutional investors at the IPO firm level. Relation_c1 has a mean of 0.357, suggesting that the institutional investors participating in a book-build pay almost 35.7 percent of the total trading revenue earned by the underwriter. On average, the underwriters market approximately 23 funds for the institutional investors as a whole during the five month period (Relation_c2). 4. Empirical Results 4.1 Relationship and Bidding Behavior In this section, we empirically test whether the bidding characteristics of related institutional investors are affected by the relationship between institutional investors and the underwriters. By estimating the logistic model of equation (2), we examine whether the likelihood of participating in a book-building is affected by the relationship between 24

27 the institutional investor and the underwriter. Columns (1) and (2) in Table 2 provide the multivariate statistical analysis results. Consistent with our hypothesis, the coefficient of Relation is 7.224, positive, and significant at the 5 percent level. An increase of one standard deviation in Relation is accompanied by a percentage point increase in log odds, indicating an increase of percentage point in the probability of participating in the book-building. The evidence suggests that when an institutional investor has a good relationship with the underwriter, the investor is more likely to participate in the book-building process and to submit bids. Using the alternative measure of the relationship provides consistent results. The coefficient of Fund_New is 0.401, positive, and significant at the 5 percent level, suggesting that institutional investors cater to the underwriter to market new funds. Notably, some of our control variables explain why institutional investors decide to participate in the book-building process. Not surprisingly, larger fund management companies (Fbig) are more likely to participate in a book-building. MKT is also positively related to the probability of an investor s involvement, implying that when the market return is good, institutional investors are more likely to participate in the book-building process. Leverage (Lev) has negative effect on the decision of institutional investors to bid. Institutional investors are less likely to participate in the book-building of highly leveraged IPO firms. We do not find significant results on DAC_D, suggesting that the decision to participate in a book-building is not affected by the earnings quality of the issuer. Columns (3) - (6) of Table 2 show the results from the Tobit model of equation 25

28 (3). The model tests the effect of the business relationship on the degree of optimism in bids. We compare the bids to the fair value estimated using information from underwriters reports and the middle point of the initial price range, respectively (Optimism1 and Optimism2). The key independent variable, Relation, has positive coefficients of and 4.473, significant at the 5% and 1% levels, respectively. An increase of one standard deviation in Relation is associated with an 8.6 percentage point increase in the bidding price over benchmarks. This indicates that a good business relationship between the underwriter and the fund management companies is associated with more optimistic bids. As the relationship becomes closer (more trading commission fees), the bids become more favorable. Fund_new is also positively and significantly related to the degree of optimism, suggesting that fund management companies who rely on the support of underwriters to market new funds around the time of a book-building submit more optimistic bids during the book-building process. Overall, the results in Table 2 are consistent with H1, which suggests that institutional investors who have a close relationship with the underwriter are more likely to participate in the bidding process and also provide more optimistic bidding prices. [INSERT TABLE 2 ABOUT HERE] 4.2 Firm Quality and Bidding Behavior Previous evidence indicates that the relationship between institutional investors and the underwriter influences the bidding behavior of institutional investors. In this 26

29 section, we further examine whether the optimistic bidding from related institutional investors is conditional on firms earnings quality. Based on the level of discretionary accruals, we separate IPO firms into two groups. If an IPO firm has higher discretionary accruals than the median of all IPO firms, we categorize the firm in the inferior quality group (DAC_D=1). IPO firms with lower discretionary accruals are categorized in the superior quality group (DAC_D=0). We estimate equations (5) and (6), which predict that α 2 and β 2 will be positive. [INSERT TABLE 3 ABOUT HERE] In Columns (1), (3), and (5) of Table 3, we find that the coefficients of Relation*DAC_D are significantly positive. The results suggest that related institutional investors bid more optimistically when the IPO firms have inferior earnings quality. Columns (2), (4), and (6) of Table 3 present the results using the alternative measure of the relationship (Fund_New). The coefficients of Fund_New*DAC_D are also significantly positive. This consistent evidence suggests that institutional investors are more likely to bid a higher price when the underwriter needs their help most, i.e., when IPO firms have low earnings quality. The results are consistent with our second hypothesis. As a robustness check, we also use a dummy variable of earnings decline that equals 1 if the earnings of the IPO firm decline after the IPO and 0 otherwise, to measure firm quality. If a firm s earnings decline after the IPO, it suggests that the 27

30 firm is more likely to have inflated earnings before the IPO and we categorize it as a firm with inferior earnings quality. The tests yield similar results to those reported in Table Optimism at the firm level In the preceding tests, we focus on the bidding behavior of institutional investors. In this section, we investigate the effect of the institutional investor-underwriter relationship on the aggregated bidding prices at the level of the IPO firm. If the bidding behavior of individual institutional investors is affected by the business relationship, we expect that the effects will also be reflected at the IPO firm level. Specifically, when more related institutional investors participate in the book-building process, the average bidding prices for the IPO firm will be more optimistic. Table 4 presents the results of the OLS regression of the degree of optimism on the institutional investor-underwriter relationship at the firm level (equation 7). Consistent with the findings documented in Table 2, the coefficients of Relation_c1 and Relation_c2 are significantly positive in Columns (1) and (3), which suggests that when more related institutional investors participate in the bidding process, the overall bidding prices are more optimistic. In addition, the coefficients of Relation_c1 and Relation_c2 are significantly larger in the inferior quality group than those in the superior quality group (Columns (2) and (4)), suggesting that the effect of the 11 For brevity, the results are not reported but are available upon request. 28