IPO Underpricing to Retain Family Control under Concentrated Ownership: Evidence from Hong Kong

Authors

  • Xin Yu,

  • Ying Zheng

    Search for more papers by this author
    • The authors are respectively from Nanjing University and Sun Yat-Sen University. They thank Martin Walker (editor) and an anonymous referee for helpful comments and suggestions. They appreciate helpful suggestions from Joseph Fan, Qiang Cheng, Xiongsheng Yang, Qingchuan Hou, Siu Kai Choy, and seminar participants at Nanjing University, participants at the 2009 International Conference on Corporate Finance and Governance in Emerging Markets. Yu gratefully acknowledges financial support from the National Natural Science Foundation of China (Grant No. 70902008). Zheng gratefully acknowledges financial support from the National Natural Science Foundation of China (Grant No. 71002059). All errors remain the authors' own.


Xin Yu, School of Business, Nanjing University, 22 Hankou Road, Nanjing, China P.R., 210093.e-mail: yuxinacy@nju.edu.cn

Abstract

Abstract:  This study examines whether family firms utilize IPO underpricing in order to retain family control as a means of avoiding outside blockholders. Using a sample of firms listed in Hong Kong where corporate ownership is concentrated, we find that larger IPO underpricing is associated with stronger family involvement, and with greater potential for diffusing ownership among family members. We further find evidence that firms with strong family involvement attract more subscription for their shares, have less concentrated outside blockholdings, and reduce ownership more slowly than do firms with weak family involvement, suggesting that family owners use underpricing to reduce outside blockholdings. We also find that firms controlled by family trusts have less IPO underpricing, suggesting that IPO underpricing and the family trust are effective substitute methods for retaining family control.

1. INTRODUCTION

This paper investigates whether family firms utilize IPO underpricing in order to retain family control as a means of avoiding outside blockholders within an institutional environment where ownership is concentrated. A substantial body of literature investigates the theoretical and empirical underpinnings of the consistent short-run underpricing of IPOs, within which a branch of the literature explains the underpricing of IPOs in terms of the agency conflicts between managers and original shareholders, under an assumption of separation of ownership and control (e.g., Aggarwal et al., 2002; Aggarwal et al., 2002; Chahine and Goergen, 2011; and Ljungqvist and Wilhelm, 2003). However, the assumption can be applied to only a limited number of markets given the recent line of corporate governance literature has shown the prevalence of concentrated firm ownership throughout the world (e.g., Anderson and Reeb, 2003; Claessens et al., 2000; and La Porta et al., 1999). Concentrated ownership with differing types of owners (e.g., family or state) is accompanied by different incentives and conflicts among the participants, rather than by conflicts between managers and original shareholders. Hence, an important but understudied issue is how to understand IPO underpricing in markets that have a prevalence of concentrated ownership. We address this issue by investigating underpricing of IPOs by family firms, and the determinants of this condition, in the context of Hong Kong, where concentrated ownership is prevalent.

The IPO literature (e.g., Biais et al., 2002; Hill and Wilson, 2006; and Ljungqvist and Wilhelm, 2003) reveals that agency conflicts among shareholders, managers, and underwriters that are due to a separation of ownership and control are an important driving factor for IPO underpricing. Because most family owners are not passive owners, but actively manage firms and preserve their control (Morck and Yeung, 2004), family firms are usually regarded as exempt from this type of agency conflict because there is little separation between ownership and control (e.g., Almeida and Wolfenzon, 2006; Johnson et al., 2000; and Morck et al., 2004). Claessens et al. (2000) show that more than eighty percent of companies in East Asia are in the charge of managers who belong to the controlling group. Thus, as firm managers, family owners should have stronger incentives to monitor underwriters and bargain for better prices, which should result in less need for IPO underpricing.

On the other hand, Brennan and Franks (1997) show that underpricing is used to ensure oversubscription and rationing in the share allocation process, in order to allow owners to discriminate among applicants and reduce the block size of new shareholders. They suggest that underpricing is a means by which to entrench managerial control by avoiding monitoring by a large outside shareholder. Family firms are unique investors because of their desire to pass firms to future generations and their concern for the firms’ long-term survival (Anderson et al., 2003; and Laeven and Levine, 2008). Given the potential for private benefits, corporate control is of great importance for family firms (Johnson et al., 2000; and Morck et al., 2004). As well, the benefits of bypassing external financing (Almeida and Wolfenzon, 2006) in an environment with imperfect investor protection support such emphasis on corporate control by the family owners. However, in many countries non-management blockholders are popular (e.g., Faccio and Lang, 2002; and Laeven and Levine, 2008), allowing a situation where corporate control may be challenged by non-management blockholders who can play an effective monitoring role, and can restrain the controlling owners’ decisions and activities. Therefore, we conjecture that IPO underpricing is utilized by family firms to structure ownership in a way that enhances family control and mitigates outsider monitoring by attracting oversubscription and diffusing outside blockholdings.

Although non-management blockholders may increase firm value (e.g., Chung and Kim, 1999; Claessens and Fan, 2002; and Lins, 2003), family firms generally do not appreciate them, because monitoring from blockholders is potentially costly. First, blockholders can access proprietary information that could be key to a family firm's success. The potential for leaking such proprietary information for instance, relationships with politicians (e.g., Fan and Wong, 2002; and Leuz and Oberholzer-Gee, 2006) is of great concern to family firms, especially in a weak institutional environment. Second, a family firm's ultimate goal is to maximize the family utility. Monitoring by blockholders could affect firms’ financing or investment decisions, which the family views as mainly serving their own long-term interests, rather than those of other shareholders. According to the Hong Kong Companies Ordinance,1 any two shareholders have the right to convene a general meeting, and investors who represent not less than one-twentieth of the total voting rights of all the members have a right to vote at general meetings. Related party transactions and other significant transactions, which are the usual methods used by family owners to absorb private benefits, have to be supported at general meetings. Thus, family firms in Hong Kong have strong incentives to enhance and safeguard the family's effective control by avoiding challenges from outside blockholders. A third reason that family firms resist outside blockholders is the threat of increased conflict with the controlling owner, and of corresponding injury to family interests. Laven and Levine (2007) find that, for firms with multiple blockholders, large shareholders are less likely to cooperate when they are different types of owners.2 Finally, ownership may be diffused among family members when controlling rights succeed to heirs, resulting in no single family member having shareholdings large enough to retain control. At that point, large outside blockholders and conflicts among family members become threats to family control. Thus, retaining family control and limiting external blockholdings are important motivations for family firms when shaping the ownership structure.

One critical problem our study faced is related to family firm research and how to define the family firm. Morck and Yeung (2004) define family firms as companies run by the heirs of those who were previously in charge or by families that are clearly in the process of transferring control to their heirs. In the IPO setting of our study, however, relatively short firm histories make it difficult to conclusively classify firms controlled by individuals as either family-owned or non-family-owned, since the time frame does not allow observation of whether the ultimate intention of the owner is to pass the ownership to the next generation or whether it is to sell the ownership. However, we do observe that the family involvement in firm operation is different in firms owned by individuals. In some firms the founder alone is on the board, while in others several family members perhaps even from two generations, are on the board and in senior management positions. We assume that when there is more involvement in the business operations by at least two generations, or by siblings or extended family (such as cousins), the increased level of family involvement in the business gives the family owners both greater ability and stronger incentive to pass firms to the next generation. Based on the level of family involvement, we classify individual-owned companies into two groups. ‘Strong group' includes firms with strong family involvement in the business, that is, two generations of the family or siblings and/or cousins are involved in the business. ‘Weak group' includes firms in which only the founding family member (or non-related founders) and their spouses are involved in the business. We first define involvement in the business as serving on the board. Given the close nexus of the Chinese family, we extend the definition of involvement to include serving as senior managers, though not on the board of directors. Even as senior managers of the company, family members are well placed to gain control, given their involvement in the day-to-day operations and decision-making processes. For the purposes of the ensuing discussion, the definition focusing solely on board representation is the ‘narrow definition' while the approach including senior managers is the ‘broad definition'. In application, we examine the change of ownership structure after an IPO, which confirms our classification of ‘strong group' versus ‘weak group’ (Section 4(v)). The descriptive statistics show that strong-group firms are more likely than weak-group firms to retain ownership, consistent with our expectation that family owners are more likely to keep ownership within the family when more family members are involved in the business.

Based on IPOs on the Main Board of the Stock Exchange of Hong Kong (SEHK), we examine difference in the level of IPO underpricing between Hong Kong firms with different levels of family involvement. Hong Kong serves as a valuable study venue because most firms in Hong Kong have concentrated ownership. Another useful aspect of Hong Kong as the study venue is the widespread use of private benefits. In papers by Brennan and Franks (1997) and Benninga et al. (2005), private benefits are important for understanding the actions of family owners and the tradeoffs between staying private and going public. Although the assumption of private benefits in Brennan and Franks (1997) has been challenged by several studies (Booth and Chua, 1996; and Zingales, 1995), the issues at the basis of the challenges are not specifically applicable to the Hong Kong market where private benefits of control have been widely documented, e.g., Ball et al. (2003), and Fan and Wong (2002). Espenlaub and Tonks (1998) find no evidence that underpricing is related to the sale of shares after the IPO in the UK and Hill (2006) finds no evidence to support the link between IPO underpricing and ownership structure of the post IPO firms with a sample of UK IPOs. However, those findings may not be generalized to markets where concentrated ownership and private benefits of control are widespread. Meanwhile, the demand for selling shares in the secondary market in Hong Kong is not as high as argued in Zingales (1995). Negotiation, rather than selling in the secondary market, is a popular method of block transfer because of the voting premium (Chung and Kim, 1999; and Nenova, 2003).

We find a higher level of IPO underpricing in firms with strong family involvement than in firms with weak family involvement. Further tests show that firms with higher potential for ownership diffusion among family members, measured by the number of family members who are in the second or third generation, or who are siblings or cousins, are more likely to undertake IPO underpricing. In addition, we find a significant positive relationship between IPO underpricing and the rate of over-subscription. We also find that the concentration of outside blockholders is the same when the IPO is offered between the two groups, but after the IPO it is lower for firms in the strong group than it is for firms in the weak group, indicating that the IPO underpricing successfully diffuses outside blockholdings. The results support the argument that the strong group uses underpricing of IPOs to retain control more than do firms in the weak group. We find a negative although insignificant association between the level of IPO underpricing and the loss of private benefits of control held by the largest owners, providing weak evidence for Benninga et al.'s (2005) theory on the tradeoff between the loss of the private benefits of control and the gains of diversification. According to Benninga et al.'s (2005) theory, the owners of family firms give up less private benefits from the control, and thus can accept a lower offering price and more substantial IPO underpricing. Our findings continue to hold when the loss of private benefits of control is contained. We also find that firms controlled by family trusts are less likely to retain control through IPO underpricing, suggesting that IPO underpricing and the family trust are effective substitute methods for retaining family control. These results indicate that for family-owned firms that go public in Hong Kong the primary motive is to obtain financing without compromising their effective control.

Our work makes several principal contributions. The literature has shown that family firms use pyramids or dual class shares to retain their control, and can do so with comparatively less cash flow (e.g., Almeida and Wolfenzon, 2006; Claessens et al., 2000; and Faccio and Lang, 2002). We show that family firms also utilize IPO underpricing to diffuse outside shareholdings, so as to avoid monitoring and to retain their effective control. Second, our study extends the literature by explaining IPO underpricing from the perspective of the issuers’ incentive. Zingales (1995) assumes that firms go public because of the value-maximizing decision made by their initial owners, whose ultimate intention is to sell the company. Ayyagari and Doidge (2010) show that controlling shareholders of foreign firms use a US cross-listing to sell control. However, those theoretical models and empirical results on issuers’ incentive in developed markets may not hold in an environment with a different instituional arrangement. Kim et al. (1993) find that IPO underpricing is lower when existing shareholders intend to diversify their holdings than when entrepreneurs view the IPO as a resort for raising funds in Korea. Our paper expands on this view by showing that the primary purpose of a family owner's going public in a market with weak investor protection is to seek additional financing, rather than to eventually sell the company. Third, the IPO literature explains IPO underpricing from the perspective of information asymmetry, but our results suggest that asymmetric information may not be the primary driver of IPO underpricing when the ownership is concentrated. Finally, our study provides evidence that promotes greater understanding of the degree of importance placed on retention of control (even after IPOs) by family firms in Asia and Europe, and in other parts of the world where markets have weak property rights protection.

The paper is organized as follows. In Section 2, we introduce the background of the study. Data and descriptive statistics are described in Section 3, empirical tests are reported in Section 4, and we summarize our study in Section 5.

2. BACKGROUND

(i) The Institutional Environment in Hong Kong

Zingales (1995) argues that a firm goes public as a result of the value-maximizing decision made by an initial owner who wants to eventually sell his company. However, the assumption of eventually selling companies through going public may not be valid in emerging Asian markets. There are large private benefits of control under the institutional environment with poor protection of property rights. Close relationships between entrepreneurial tycoons and politicians also generate private benefits of control through political rent-seeking activities (Ball et al., 2003; and Fan and Wong, 2002). Shleifer and Wolfenzon (2002) believe that this kind of institutional environment prevents large shareholders from selling out, because it is neither an acceptable nor a viable alternative for them to assume the role of a diversified passive investor. Their point of view is consistent with that of Morck and Yeung (2004), who propose that control pyramids greatly magnify the private benefits of control from rent-seeking activities. The prevalence of family-controlled firms in Hong Kong is widely documented (e.g., Claessens et al., 2000; and La Porta et al., 1999). Fan and Wong (2002) argue that concentrated ownership is endogenously determined by the institutional environment in order to keep proprietary information and specific human capital in East Asia. Ball et al. (2003) explained that family-controlled businesses and guanxi networks in the region are partially determined by the significance of the family in traditional Chinese ideology, giving rise to a system of personal networks that revolves around informal relationships, rather than formal legal contracts. These studies implicitly recognized Hong Kong as a jurisdiction with strong legal protection, and explained the importance of family control based on factors other than legal protection.

However, although Hong Kong is consistently cited for its commitment to free-market principles, it has been criticized for the perpetuation of its land ownership policy, which led the Financial Times to contend that ‘the government's monopoly on land sales preserves its incentive to collude with property tycoons to push up property prices and magnifies swings in the economic cycle.'3 The close connection between the government and business, a relationship founded primarily on Hong Kong's peculiar system of land ownership and property development, has been documented by Goodstadt (2005). Additional anecdotal evidence of government intervention has highlighted the degree of importance that family owners place on tight corporate control, which, in turn, keeps proprietary information out of the public domain in Hong Kong.

(ii) Types of Listings in Hong Kong

The types of listings on the SEHK include an offer for subscription of shares by the company, an offer for sale of shares by existing shareholders, a placement, and an introduction. According to the listing rules of SEHK, the definitions of the types of listing are as follows,

  • • Offer for subscription– an offer to the public of new issued securities for subscription.
  • • Offer for sale– an offer to the public of securities already in issue or agreed to be subscribed, usually held by existing shareholders.
  • • Placing – obtaining of subscriptions for or the sale of securities to persons selected or approved by the issuer or intermediary.
  • • Introduction– an application for listing of securities already in issue where no marketing arrangements are required. This method is usually applied when a company is already listed on another market, or its shares are being migrated from one exchange to another, or its securities are issued in exchange for those of listed issuers. Examples include a dual listing in SEHK, and moving to the main board from Growth Enterprise Market (GEM) board.4

SEHK rules mandate that there be a fair allotment process in both the offer for subscription and the offer for sale so that every investor who applies at the same price for the same number of securities receives equal treatment.

Both fixed pricing and book-building are used in Hong Kong in IPO pricing. Listings can be solely by placement, provided that a minimum prescribed percentage of the listed securities are held at all times by independent members of the public. However, the SEHK generally does not permit a new applicant to be listed solely by way of a placing if there is likely to be significant public demand for the securities. In practice, placing is usually done in tandem with offers for subscription and/or sale of shares. When an IPO includes both a placing tranche and a public subscription tranche, book-building is used in IPO pricing only with institutional investors participating in the placing tranche (Morales-Camargo, 2007). The minimum allocation of shares for the public offering is normally set at an initial level of ten percent of the shares offered during the IPO, subject to the ‘claw-back' mechanism,5 which operates to increase the number of shares available under the public subscription tranche to 30%, 40% and 50% when the total demand for shares in this tranche is over-subscribed by at least 15 times, 50 times or 100 times, respectively. As a result of the potential trigger of the claw-back mechanism, the placing in tandem with offers for subscription and/or sale can be regarded as fixed-price offers with fair allotment in some degree, even when the book-building method is used.

3. DATA AND DESCRIPTIVE STATISTICS

Our sample covers the IPOs on the Main Board of the SEHK over a five-year period from 2002 to 2006. With the exception of data on stock prices and the Hang Seng Index, which are obtained from Thomson Financial, and industry classification, which is obtained from the WIND database produced by Shanghai Wind Information Company, all items are hand-collected from prospectuses, IPO allotment results, and the new listing reports from the websites of the SEHK and Clearing Limited.

We begin with the full population of 269 IPOs on the Main Board of the SEHK from 2002 to 2006 as summarized in the annual ‘New Listing Report.’ We eliminate the 21 observations of listings by introduction, 5 observations that involve the listings of investment trusts or investment companies, 16 observations that have been delisted or acquired, and 11 observations that have none of the information needed for the tests. The remaining sample has 216 observations, of which 48 are H-shares and 17 are Red Chips. The classification of H-shares and Red Chip firms follows that prescribed by the SEHK. H-shares refer to shares of companies that are incorporated in Mainland China. Red Chip stocks refer to companies that are incorporated outside China and at least 30% are directly held by Mainland Chinese entities and/or through companies that are controlled by such entities. Mainland China entities include state-owned organizations and entities controlled by provincial or municipal authorities. Hence, Red Chip companies are essentially state-owned companies. Since the focus of our paper is on family firms, a population which is derived from the set of individually owned companies, we exclude the state-owned companies (Red Chips) from our sample. We also exclude H-share companies that are incorporated in Mainland China. The final sample contains 151 IPOs in Hong Kong, presented in Table 1.

Table 1. 
Sample Selection
Sample Selection  
  1. Notes:
    This table describes the sample seletion process. The final sample contains 151 IPOs.

Original sample: IPOs from 2003 to 2006269
Minus: 
 By introduction(21)
 Investment firms/trusts (5)
 Delisted or acquired(16)
 Related information missing(11)
Total Sample216
State-owned companies(65)
 H-shares48
 Red Chip17
Individual-owned sample151

The year distribution and industry distribution of the sample are presented in Table 2. The number of IPOs in each year is increasing in the sample period, except in year 2006. However, the sample observations are not equally distributed in all industries. The three industries Industrial Products, Textiles, and Household Appliances have the most observations, while Mining, Agricultural Products, and Petroleum and Gas have the least.

Table 2. 
Distribution of IPO Sample in Hong Kong
Panel A: By Year
Year Number Percentage
  1. Notes: This table presents distribution of IPO sample over years as well as across industries. Strong group includes firms with two generations of the family or siblings and/or cousins involved in the business. Weak group includes firms in which only the founding family member (or non-related founders) and their spouses are involved in the business. The narrow definition refers to family involvement defined based on the family relationship among board members while the broad definition refers to family involvement defined based on the family relationship among board members and senior managers.

2002 2315.23
2003 2516.56
2004 3120.53
2005 3925.83
2006 3321.85
Total151100
Narrow Definition Strong Weak
YearNumberNumber
2002 617
20031213
2004 823
20051524
20061518
Total5695
Broad Definition Strong Weak
YearNumberNumber
2002 617
20031312
20041021
20051722
20061716
Total6388
Panel B: By Industry
Industry Number Percentage
Semiconductors  31.96
Health Care Products  85.23
Mine  10.65
Publishing & Printing  31.96
Real Estate  85.88
Telecommunications  21.31
Textiles 1912.42 
Industrial Products 2516.34 
Household Appliances 1811.76 
Construction  31.96
Metal  31.96
Hotel & Entertainment  31.96
Retail 106.54
Agricultural Products  10.65
Automobiles  31.96
Software & Related Services  75.23
Petroleum & Gas  10.65
Food  74.58
Raw Materials  95.88
Transportation  31.96
Support Services  63.92
IT Hardware  85.23
Total 151100

Table 3 describes the IPO underpricing for the sample firms classified in the narrow definition. IPO underpricing is measured by both the first-day return (FDR), which is equal to the closing market price on the first day less the offer price, divided by the offer price, and the market-adjusted first-day return (Adj. FDR), which is calculated as the first-day return minus the market return in the same day. The market return is measured by the return of the Hang Seng Index. The three left-hand columns present FDR, while Adj. FDR is presented in the middle. The mean of Adj. FDR ranges from a low of 3.18% in 2005 to a high of 18.16% in 2006, so the annual variation in underpricing is large. The mean and median of IPO underpricing of firms in the strong group are larger than those of firms in the weak group in the pool and in most years. The three right-hand columns show the money left on the table during the IPO process. From 2002 to 2006, on average, firms in the SEHK left HK$50.64 million on the table during the IPO process; firms in the strong group left HK$90.85 million, while firms in the weak group left HK$26.09 million. There is no significant difference in results when the sample is classified using the broad definition.

Table 3. 
Summary of IPO Underpricing (Narrow Definition)
First-Day Return (%) Adjusted First-Day Return (%) Money on the Table (HK$M)
Year Strong Weak Total Strong Weak Total Strong Weak Total
  1. Notes: This table presents the mean and median of First-day return (FDR) and Money on the table (HK$M) for the IPO sample. The narrow definition refers to family involvement defined based on the family relationship among board members. Strong group includes firms with two generations of the family or siblings and/or cousins involved in the business. Weak group includes firms in which only the founding family member (or non-related founders) and their spouses are involved in the business. First-day return (FDR) is measured as the difference between the closing price on the offer date and the offer price, divided by the offer price. Adjusted First-day return (Adj. FDR) is measured as the First-day return minus the Hang Seng Index return on the same day. Money on the table is measured as the product of the number of shares issued and the difference between the closing price on the offer date and the offer price. Variables’ medians are reported in parentheses.

200210.94 9.229.6710.54 9.739.949.558.548.80
 (9.64)(5.88)(9.09)(9.05)(6.84)(8.68)(9.55)(3.00)(7.28)
20039.38−1.96  3.489.53−2.48  3.287.30−11.17   −2.31  
 (6.12)(3.75)(3.75)(5.56)(2.68)(3.21)(5.77)(2.16)(2.16)
2004−1.04  6.174.31−0.83  6.294.454.0120.51 16.25 
 (0.00)(1.87)(1.12)(0.52)(2.02)(1.73)(0.00)(2.00)(1.20)
20059.09−0.32  3.308.99−0.44  3.1889.66 −2.49  32.96 
 (0.83)(0.00)(0.71)(0.69)(−0.06)  (0.45)(2.60)(0.00)(0.75)
200617.35 18.00 18.08 17.58 18.66 18.16 239.27  83.06 154.06  
 (14.47) (18.69) (13.49) (15.75) (12.42) (13.54) (81.60) (56.50) (67.00) 
Total10.12 6.347.7410.17 6.357.7691.27 20.07 46.48 
 (5.66)(3.45)(3.64)(5.59)(2.92)(3.37)(9.55)(3.00)(4.00)

Corporate boards of directors have an average of eight directors, including three independent non-executive or outside directors (Table 4). The size of the board and the number of independent non-executive directors are similar between the strong group and the weak group; firms in the strong group have an average of three directors with substantial shareholdings, while firms in the weak group have only two. The disclosure requirements of the Hong Kong exchange define a substantial shareholding as holdings of 5% or more of the issued share capital of the company, and mandate that there be at least three independent non-executive directors.

Table 4. 
Summary of Board Characteristics
  No. Mean Median Std.Dev. Min. Max.
  1. Notes: This table describes board characteristics for the IPO sample, by both the narrow definition and broad definition. No. of directors with substantial shareholdings is measured as the number of board directors with substantial shareholdings, defined as holdings of 5% or more of the issued share capital of the company. No. of directors is the number of total board members. No. of independent non-executive directors is the number of board directors without management positions. Strong group includes firms with two generations of the family or siblings and/or cousins involved in the business. Weak group includes firms in which only the founding family member (or non-related founders) and their spouses are involved in the business. The narrow definition refers to family involvement defined based on the family relationship among board members while the broad definition refers to family involvement defined based on the family relationship among board members and senior managers.

Full Sample
No. of directors with substantial shareholdings1512.2321.390 8
No. of directors1517.7971.82513
No. of independent non-executive directors1512.9130.782 6
Narrow Definition
Strong group       
No. of directors with substantial shareholdings563.0531.371 8
No. of directors567.9172.10513
No. of independent non-executive directors562.9130.692 5
Weak group       
No. of directors with substantial shareholdings951.7521.170 7
No. of directors957.7381.65512
No. of independent non-executive directors952.9230.832 6
Broad Definition
Strong group       
No. of directors with substantial shareholdings632.9431.331 8
No. of directors637.8472.02513
No. of independent non-executive directors632.9230.682 5
Weak group       
No. of directors with substantial shareholdings881.7321.210 7
No. of directors887.7681.68512
No. of independent non-executive directors882.9130.852 6

Table 5 displays the descriptive statistics for the variables used in the regression tests. The mean and median of FDR (and Adj. FDR) of firms in the strong group are larger than those of firms in the weak group, and the ownership retained by substantial shareholders (Retained) is larger in the strong group than in the weak group. Firms with strong family involvement have longer histories than firms with weak family involvement at the IPO date; on average, firms in the strong group have 18 years of history since incorporation, while firms in the weak group have only 12. The offer size of firms in the strong group is larger than that of firms in the weak group. The mean and median returns on sales (ROS) in the year prior to the IPO year are almost the same in the two groups. The debt level, measured by total debt/total assets, is slightly lower in the strong group than in the weak group. The level of separation between cash flow rights and voting rights (CV) is 91% on average, and there is no difference between the two groups.

Table 5. 
Descriptive Statistics of Key Variables
Full Sample Number Mean Median Std. Dev. Min. Max.
  1. Notes: This table provides the summary statistics of of key variables. FDR, first-day return, is measured as the difference between the closing price on the offer date and the offer price, divided by the offer price. Adj. FDR is FDR minus the Hang Seng Index return on the same day. Retained is the percentage of ownership retained by substantial shareholders. Sold is the percentage of ownership sold by insider owners in the IPO process. Commitment is a dummy variable equal to 1 if the firms use firm-commitment offering, and zero if the firms use best-effort offering. Fixedprice is a dummy variable that is equal to 1 if the issuers use the fixed pricing method, and zero if the issuers use the book-building method. Sponsor is a dummy variable equal to 1 if the sponsor or one of the sponsors is on the ranking list and zero otherwise. Auditor is a dummy variable equal to 1 if the financial statements are audited by a Big Four audit firm and zero otherwise. OS is the offering size, measured as the number of shares offered times the offer price. Age is the number of years from the year of incorporation to the date of the IPO. ROS is measured as net income divided by sales in the year prior to the IPO. Debt is measured as total debt divided by total assets in the year prior to the IPO. CV is the cash flow rights divided by the voting rights of the majority owner at the time of the IPO. Strong group includes firms with two generations of the family or siblings and/or cousins involved in the business. Weak group includes firms in which only the founding family member (or non-related founders) and their spouses are involved in the business. The narrow definition refers to family involvement defined based on the family relationship among board members while the broad definition refers to family involvement defined based on the family relationship among board members and senior managers.

FDR(%)1517.743.6717.73 −66.67   63.64 
Adj. FDR (%)1517.763.3717.68 −67.05   63.56 
Retained1510.620.690.140.100.79
Sold1510.120.000.180.001.00
Commitment1510.931.000.260.001.00
Fixedprice1510.390.000.490.001.00
Sponsor1510.330.000.470.001.00
Auditor1510.881.000.320.001.00
OS (HK$M)151544177984117,050
Age15114.56 12.00 9.594.0059.00 
ROS1510.190.150.130.010.82
Debt1510.160.130.160.000.69
CV1510.911.000.180.041.00
Narrow Definition Number Mean Median Std. Dev. Min. Max.
Strong group       
FDR (%)5610.12 5.6615.05 −20.93   44.00 
Adj. FDR (%)5610.17 5.5915.04 −20.83   46.14 
Retained560.660.710.120.330.79
Sold560.140.000.220.001.00
Commitment560.931.000.260.001.00
Fixedprice560.380.000.490.001.00
Sponsor560.410.000.490.001.00
Auditor560.911.000.280.001.00
OS (HK$M)566812261120115,800
Age5618.21 15.50 11.79 4.0059.00 
ROS560.190.160.130.010.77
Debt560.150.130.150.000.53
CV560.911.000.180.341.00
Weak group       
FDR (%)956.343.4119.07 −66.67   63.64 
Adj. FDR (%)956.352.9219.00 −67.05   63.56 
Retained950.610.670.150.100.75
Sold950.110.000.160.000.75
Commitment950.931.000.260.001.00
Fixedprice950.400.000.490.001.00
Sponsor950.280.000.450.001.00
Auditor950.861.000.350.001.00
OS (HK$M)95463151893507,050
Age9512.41 11.00 7.274.0043.00 
ROS950.190.150.140.010.82
Debt950.170.140.160.000.69
CV950.911.000.190.041.00
Strong group       
FDR (%)6310.42 5.7114.69 −20.93   44.00 
Adj. FDR (%)6310.46 5.6214.66 −20.83   46.14 
Retained630.660.710.120.330.79
Sold630.140.000.210.001.00
Commitment630.941.000.250.001.00
Fixedprice630.400.000.490.001.00
Sponsor630.400.000.490.001.00
Auditor630.891.000.320.001.00
OS (HK$M)636232021,070115,800
Age6317.76 14.00 11.59 4.0059.00 
ROS630.190.160.130.010.77
Debt630.150.130.140.000.53
CV630.921.000.170.341.00
Weak group       
FDR (%)885.822.9119.47 −66.67   63.64 
Adj. FDR (%)885.842.7219.41 −67.05   63.56 
Retained880.600.650.160.100.75
Sold880.110.000.160.000.75
Commitment880.921.000.270.001.00
Fixedprice880.390.000.490.001.00
Sponsor880.280.000.450.001.00
Auditor880.881.000.330.001.00
OS (HK$M)88487164923507,050
Age8812.27 11.00 7.094.0043.00 
ROS880.190.150.140.010.82
Debt880.170.140.170.000.69
CV880.911.000.190.041.00

In aggregate, 33% of the sample firms chose one of the Top 15 investment banks as sponsors for their IPOs, and 88% of the sample firms chose one of the Big Four accounting firms. The percentage of firms in the strong group that chose a Top 15 investment bank for a sponsor is 41% (40% under the broad definition), compared to 28% of firms in the weak group, while the percentage choosing Big Four accounting firms is similar for both groups. Overall, 93% (7%) of the sample firms use firm-commitment issue (best-effort issue), with no difference between the two groups. As to the IPO pricing mechanism, 39% of firms use fixed pricing, while 61% of firms use book-building, a result that is consistent with Morales-Camargo (2007). The percentage is similar between the two groups.

4. EMPIRICAL RESULTS

(i) Basic Regression Tests

To test the question of whether family control incentives enlarge IPO underpricing in Hong Kong, we regress a firm's IPO underpricing on the degree of family involvement in the business and applicable control variables. Following Butler et al. (2009), we control for offering characteristics, IPO pricing mechanism, and other firm characteristics:

image(1)

Following Loughran and Ritter (2002 and 2004), and Butler et al. (2009), we measure underpricing by initial return in the first trading day (FDR), which is equal to the closing market price on the first day less the offer price, divided by the offer price. To control for the market volatility effect, underpricing is also measured by the market-adjusted first-day return (Adj. FDR), which is defined as the first-day return minus the Hang Seng Index return on the same window. Following Brennan and Franks (1997), we also use return in five trading days the closing market price on the fifth trading day minus the offer price, divided by the offer price to avoid the possibility that the first–day returns are manipulated. All results are qualitatively the same.6

Family is a dummy variable that is equal to 1 if the firm has strong family involvement (the strong group) and 0 otherwise (the weak group). We expect that the coefficient on Family should be positive because firms with strong family involvement have stronger incentive to retain control through IPO underpricing than firms with weak family involvement.

Retained is the percentage of ownership retained by substantial shareholders. It is difficult to predict the relationship between Underpricing and Retained. On the one hand, increased ownership concentration lowers liquidity (Heflin and Shaw, 2000). If the largest owners retain too many shares, they may become concerned about market liquidity and use underpricing to diffuse outside owners and increase market liquidity (Booth and Chua, 1996). On the other hand, largest owners retaining more shares are more likely to entrench in management or expropriate minority shareholders due to lack of monitoring. Fewer outside investors are willing to purchase the shares, and then the level of underprcing is likely to be lower.

Sold is the percentage of ownership sold by insider owners during the IPO. Leung and Menyah (2006) suggest that selling only a small proportion, or none, of pre-IPO shares greatly lowers the issuer's underpricing cost. If initial owners sell a large proportion of pre-IPO shares during the IPO, they will hesitate to underprice much. We therefore expect that Underpricing and Sold are negatively related.

Commitment is a dummy variable that is equal to 1 if the firm uses a firm-commitment offering, and zero if the firm uses a best-effort offering. Firm-commitment offering is an arrangement in which an underwriter buys the issue from the issuers and guarantees sale of shares to investors while best-effort offering is an arrangement in which an investment bank agrees to do its best to sell the offering to the public but does not buy the securities. We expect a positive coefficient on Commitment because underpricing makes it easier for underwriters to sell out shares.

Fixedprice is a dummy variable for the IPO pricing method that is equal to 1 if the issuer uses the fixed pricing method, and zero if the issuer uses the book-building method. It is difficult to predict the relationship between Underpricing and Fixedprice.

Sponsor is a dummy variable for the sponsor(s)’ ranking, as extracted from ‘Asia (Ex-Japan) Equity Sales Rankings,’ an annual ranking covering the Top 15 investment banks based on equity sales that is published by Institutional Investor. This variable equals 1 if the sponsor or one of the sponsors is on the ranking list, and 0 otherwise. A negative coefficient on Sponsor is expected, since we expect sponsors with better reputations to be less likely to underprice (Beatty and Ritter 1986; and Carter and Manaster, 1990).

Auditor is a dummy variable equal to 1 if the financial statements are audited by Big Four firms (or their affiliates), and zero otherwise. We expect the sign of Auditor to be negative because Big Four auditors signal the quality of the issue and reduce underpricing.

Two variables, LogOS and Log (1+Age), are proxies for valuation uncertainty. Age is the number of years from the year of incorporation to the date of the IPO. The data for the year of incorporation are hand-collected from the prospectuses and, where possible, the year of original incorporation is used, rather than the reincorporation year, when the firm was reorganized in preparation for listing. Log (1+Age) is the natural log of 1 plus the number of years from the year of incorporation to the date of the IPO. LogOS is the natural log of the offering size, calculated by multiplying the number of shares offered at the offer price. We expect these two coefficients to be negative because prior studies (Beatty and Ritter, 1986; and Booth and Chua, 1996; and Ljungqvist and Wilhelm, 2003) documented the comparative ease of valuing larger IPOs and IPOs of firms with longer histories.

ROS is measured as net income divided by sales in the year prior to the IPO. Debt is measured as total debt divided by total assets in the year prior to the IPO. These two variables are included to control for firm characteristics. We expect the coefficients of these two variables to be negative because firms with better performance (ROS) and lower debt level (Debt) will be less underpriced.

CV is measured as the cash-flow rights divided by the voting rights of the ultimate owner at the time of the IPO. This variable is added in order to control for agency cost; if the controlling owners have high voting rights but low cash-flow rights, they may have incentive to expropriate outside shareholders. Other large investors may ask for a discount in the offering price as an offset to potential expropriation. We therefore expect the coefficient of CV to be positive.

Finally, we use industry dummies to control for industry fixed effects and year dummies to control for year fixed effects in the regressions.7

Table 6 presents the results of the regression analysis with four versions of the regression model estimated separately. The first two regression models are based on the narrow definition of family involvement, and the other models are based on the broad definition. The dependent variables in models (1) and (3) are the raw first-day return (FDR), while the dependent variables in model (2) and (4) are the market-adjusted first-day return (Adj. FDR). The White-t standard errors are utilized to adjust for heteroskedasticity. In all the regression models there is a positive relationship between the family involvement dummy and the level of IPO underpricing, significant at least at the 5% level. This result is consistent with our prediction that firms in the strong group are more likely to underprice during IPOs than are firms in the weak group.

Table 6. 
Regression Estimates of the Relationship between IPO Underpricing and Family Involvement
Variable Pred. Sign Narrow Definition Broad Definition
(1) FDR (2) Adj. FDR (3) FDR (4) Adj. FDR
  1. Notes: This table presents the regression results estimating the relation between IPO underpricing and issuers’ characteristics. The dependent variable is IPO underpricing measured by FDR and Adj. FDR. Family is a dummy variable equal to 1 if the firm is in the strong group, and 0 if the firm is in the weak group. Strong group includes firms with two generations of the family or siblings and/or cousins involved in the business. Weak group includes firms in which only the founding family member (or non-related founders) and their spouses are involved in the business. The narrow definition refers to family involvement defined based on the family relationship among board members while the broad definition refers to family involvement defined based on the family relationship among board members and senior managers. Retained is the percentage of ownership retained by substantial shareholders. Sold is the percentage of ownership sold by insider owners in the IPO process. Commitment is a dummy variable equal to 1 if the firms use firm-commitment offering, and zero if the firms use best-effort offering. Fixedprice is a dummy variable that is equal to 1 if the issuers use the fixed pricing method, and zero if the issuers use the book-building method. Sponsor is a dummy variable equal to 1 if the sponsor or one of the sponsors is on the ranking list and zero otherwise. Auditor is a dummy variable equal to 1 if the financial statements are audited by a Big Four audit firm and zero otherwise. OS is the offering size, measured as the number of shares offered times the offer price. Age is the number of years from the year of incorporation to the date of the IPO. ROS is measured as net income divided by sales in the year prior to the IPO. Debt is measured as total debt divided by total assets in the year prior to the IPO. CV is the cash flow rights divided by the voting rights of the majority owner at the time of the IPO. The heteroskedasticity adjusted White-t statistics are in parentheses. *significant at the 10% level; **significant at the 5% level; ***significant at the 1% level.

Intercept −27.369−29.227−30.169−32.182
  (−0.896)(−0.973)(−1.042)(−1.131)
Family+7.356**7.400**8.255**8.209**
  (1.999)(2.011)(2.480)(2.467)
Retained+/−−17.336−17.995*−18.985*−19.613*
  (−1.579)(−1.660)(−1.735)(−1.817)
Sold-−7.691−7.945−7.303−7.554
  (−1.217)(−1.292)(−1.156)(−1.228)
Commitment+13.048***13.079***12.680***12.708***
  (3.088)(3.127)(3.074)(3.116)
Fixedprice+/−0.4110.5840.0680.257
  (0.110)(0.157)(0.019)(0.071)
Sponsor-−6.118−6.207*−6.408*−6.479*
  (−1.630)(−1.665)(−1.730)(−1.761)
Auditor-−7.116*−6.952*−6.777−6.608
  (−1.660)(−1.661)(−1.624)(−1.619)
LogOS-3.319**3.386**3.592***3.663***
  (2.518)(2.609)(2.895)(2.999)
Log(1+Age)-−10.919**−10.942**−10.744**−10.749**
  (−2.473)(−2.470)(−2.516)(−2.509)
ROS-−36.810***−37.274***−36.602***−37.018***
  (−3.708)(−3.780)(−3.837)(−3.910)
Debt-−15.275−15.422−13.759−13.938
  (−1.540)(−1.576)(−1.385)(−1.422)
CV+7.2137.4455.9916.224
  (1.097)(1.125)(0.910)(0.940)
Industry&year dummies YesYesYesYes
No. of observations 151151151151
Adj.R2 0.4380.4440.4470.452

Most of the remaining control variables are significant and their coefficients have consistent signs as predicted. For example, Commitment is positively correlated with Underpricing, showing that firm-commitment offering is associated with larger IPO underpricing. Fixedprice is not significantly correlated with Underpricing,indicating that choosing either a fixed pricing or a book-building mechanism does not affect IPO underpricing. This may be due to the fact that the ‘claw-back' mechanism in book-buiding methods actually reduces the difference between fixed pricing and book-building. The perceived quality of sponsors and auditors is negatively associated with first-day return, consistent with our expectation. As we expected, there is a significantly negative association between firm age and IPO underpricing. ROS is negatively associated with the first-day return. But there is one exception. Offer size (LogOS) is significantly positively related to underpricing in all the regression models, contrary to our expectation. We leave this issue to future research.

(ii) The Potential of Ownership Diffusion Among Family Members

The founder usually has strong controlling power in the firm because of a large percentage of ownership and significant personal influence. However, when controlling rights succeed to heirs, ownership can be diffused among family members. Heirs usually have less personal influence in the firm and on the family than the founder did. Even if the family, overall, owns more than half of the shares, no single family member may have shareholdings large enough to retain control, and conflicts among family members may arise. At that point, large outside blockholders become a threat to family control.

For instance, suppose a family owns 60% shareholdings at the time of an IPO. A blockholder who has a 10% shareholding may not be a major threat to the family as a whole, but if the founder has three heirs and each heir owns 20% of the shares, an outside blockholder with 10% of the shares may challenge the family's control if there are conflicts among the heirs. Thus, we predict that firms with a greater potential for ownership diffusion among family members have a correspondingly greater incentive to avoid outside blockholders and, thus, are more likely to underprice an IPO.

We substitute the number of siblings and cousins for the dummy variable, Family, in equation (1). The results are presented in Table 7. Familymember measures the number of second- and/or third-generation family members, siblings, and cousins involved in business and proxies for the potential of ownership diffusion among family members. Panel A shows the descriptive statistics of the variable Familymember. On average, there are two family members involved in the business, excluding the founder and the founding couple. Familymember varies from 1 to 4, indicating that there is variation in the potential for ownership diffusion among family members. The results in Panel B show a positive association between Familymember and IPO underpricing in all the regressions, significant at least 5%, consistent with our expectation that firms with a larger potential for ownership diffusion among family members are more likely to underprice an IPO to avoid outside blockholdings.

Table 7. 
Regression Estimates of the Relationship between IPO Underpricing and Family Members
Panel A: Descriptive Statistics for Familymember in the Strong Group
  Number Mean Median Std.Dev. Min. Max.
Narrow definition562.1420.9214
Broad definition632.3721.1114
Panel B: Regression Estimates
Variable Narrow Definition Broad Definition
(1) FDR (2) Adj. FDR (3) FDR (4) Adj. FDR
  1. Notes: This table presents the regression results estimating the relationship between IPO underpricing and issuers’ characteristics. The dependent variable is IPO underpricing measured by FDR and Adj. FDR. Familymember is the number of the second and/or third generation, or siblings and cousins, involved in business. The narrow definition refers to family involvement defined based on the family relationship among board members while the broad definition refers to family involvement defined based on the family relationship among board members and senior managers. Retained is the percentage of ownership retained by substantial shareholders. Sold is the percentage of ownership sold by insider owners in the IPO process. Commitment is a dummy variable equal to 1 if the firms use firm-commitment offering, and zero if the firms use best-effort offering. Fixedprice is a dummy variable that is equal to 1 if the issuers use the fixed pricing method, and zero if the issuers use the book-building method. Sponsor is a dummy variable equal to 1 if the sponsor or one of the sponsors is on the ranking list and zero otherwise. Auditor is a dummy variable equal to 1 if the financial statements are audited by a Big Four audit firm and zero otherwise. OS is the offering size, measured as the number of shares offered times the offer price. Age is the number of years from the year of incorporation to the date of the IPO. ROS is measured as net income divided by sales in the year prior to the IPO. Debt is measured as total debt divided by total assets in the year prior to the IPO. CV is the cash flow rights divided by the voting rights of the majority owner at the time of the IPO. The heteroskedasticity adjusted White-t statistics are in parentheses. *significant at the 10% level; **significant at the 5% level; ***significant at the 1% level.

Intercept−31.886−33.855−32.569−34.875
 (−0.977)(−1.054)(−1.022)(−1.112)
Familymember2.965**2.984**2.942**2.890**
 (1.988)(1.997)(2.191)(2.149)
Retained−17.991−18.657*−19.510*−20.091*
 (−1.637)(−1.720)(−1.803)(−1.881)
Sold−8.093−8.349−7.406−7.655
 (−1.186)(−1.256)(−1.110)(−1.180)
Commitment13.234***13.277***13.176***13.217***
 (3.285)(3.324)(3.317)(3.349)
Fixedprice−0.0620.101−0.1010.094
 (−0.016)(0.027)(−0.027)(0.025)
Sponsor−6.959*−7.064*−7.176*−7.238*
 (−1.847)(−1.890)(−1.938)(−1.965)
Auditor−7.470*−7.310*−7.177*−6.999*
 (−1.725)(−1.729)(−1.707)(−1.703)
LogOS3.399**3.470**3.552**3.632***
 (2.418)(2.503)(2.614)(2.713)
Log(1+Age)−10.321**−10.334**−10.699**−10.671**
 (−2.293)(−2.291)(−2.370)(−2.353)
ROS−37.091***−37.585***−38.534***−38.930***
 (−3.989)(−4.078)(−4.046)(−4.122)
Debt−14.635−14.781−12.826−13.071
 (−1.489)(−1.524)(−1.294)(−1.334)
CV9.1759.4338.6278.850
 (1.397)(1.433)(1.316)(1.347)
Industry&year dummiesYesYesYesYes
No. of observations151151151151
Adj.R20.4380.4450.4450.450

(iii) Does the Underpricing of an IPO Attract More Subscription?

As mentioned in Section 2(ii), the minimum allocation of shares for the public offering (for subscription and/or sale) is normally set at an initial level of ten percent of the shares offered during the IPO, and ninety percent of the shares are offered by placing. The number of subscribers for shares offered by placing is usually much smaller than that for the public-offering shares, but the amount of shares each investor subscribes in placing is much larger than that in public offering. Although the issuer can select to whom securities are sold to in placing, the outside ownership is more concentrated than if securities are equally offered to investors in public offering. Because of the claw-back mechanism which operates to increase the number of shares available under the public-offer tranche to 30%, 40% and 50% when the total demand for shares in this tranche is over-subscribed by at least 15 times, 50 times or 100 times, respectively, over-subscription will increase the number of shares available under the public-offer tranche, which will largely diffuse outside ownership and reduce monitoring from outside ownership to family control.

In this section, we examine whether underpricing attracts more subscription. Firms are required to report the precise rate of subscription for the public-offer tranche of their IPOs, but the rate for the placing is generally not reported as a specific amount; instead, firms use terms such as ‘fully-subscribed,' ‘over-subscribed,' and ‘significantly over-subscribed.' Although evidence suggests that the rates of over-subscription for the public-offer tranche and the placing tranche are usually positively correlated, we study only the former, since these are the rates provided by the companies.

We find that the Pearson correlation between FDR (Adj. FDR) and subscription rate is 0.390 (0.393), while the Spearman correlation between FDR (Adj. FDR) and subscription rate is 0.452 (0.447), both of which are significant at 0.001. The results in Table 8 show that FDR (Adj. FDR) are still highly correlated with the subscription rate after controlling for industry and year effects, suggesting that underpricing is an effective means by which to attract more subscription and diffuse outside ownership.

Table 8. 
Regression Estimates of the Relationship between Subscription Rate on IPO Issuing and IPO Underpricing
  (1) FDR (2) Adj. FDR
  1. Notes: This table presents the regression results estimating the relationship between IPO underpricing and IPO subscription rate by small investors. The dependent variable is IPO underpricing measured by FDR and Adj. FDR. *significant at the 10% level; **significant at the 5% level; ***significant at the 1% level.

Subscription rate 2.400*** 2.460***
 (3.20) (3.21)
Industry&year dummiesYesYes
No. of observations151151
Adj-R20.338  0.341  

(iv) Does the Underpricing of an IPO Diffuse Outside Blockholdings?

We argue that IPO underpricing is a mechanism through which family owners can enhance control and reduce outside blockholdings. In this section, we investigate the ownership concentration of outside blockholdings in order to determine whether underpricing diffuses them. Following prior literature (e.g., Boubakri et al., 2005), we use the Herfindahl index (the sum of the squared holding proportion of the largest n shareholders) to measure ownership concentration. Since we focus on the concentration of outside blockholdings, we exclude the holding proportion of the controlling owner. If several shareholders come from the same family, we total their shareholdings and treat them as the controlling owner. The Herfindahl index for the top three shareholders (H3) is defined as the sum of the squared holding proportion of the second- and the third-largest shareholders. The Herfindahl index for the top five shareholders (H5) is defined as the sum of the squared holding proportion of the second- to the fifth-largest shareholders. The results are presented in Table 9.

Table 9. 
Ownership Concentration, Excluding the Controlling Owner/Family
Panel A: Narrow Definition
   IPO IPO Year End 2007
  1. Notes: This table presents the ownership concentration index, excluding the controlling owner/family at the time of an IPO, at the end of the IPO year, and at the end of 2007. The Herfindahl index (H3) is defined as the sum of the squared holding proportion of the second- and the third-largest shareholders. The Herfindahl index (H5) is defined as the sum of the squared holding proportion of the second- to the fifth-largest shareholders. Strong group includes firms with two generations of the family or siblings and/or cousins involved in the business. Weak group includes firms in which only the founding family member (or non-related founders) and their spouses are involved in the business. The narrow definition refers to family involvement defined based on the family relationship among board members while the broad definition refers to family involvement defined based on the family relationship among board members and senior managers. The mean difference is tested using the t-test, while the median difference is tested by the Wilcoxon rank sum test. *significant at the 10% level; **significant at the 5% level; ***significant at the 1% level.

Herfindahl index (H3)     
Strong groupMean0.0050.0060.008
 Median0.0000.0000.003
Weak groupMean0.0160.0160.019
 Median0.0000.0030.008
DifferenceMean−0.011−0.010−0.010
 (t-value)(−2.634)***(−2.449)***(−2.570)***
 Median0.000−0.003−0.005
 (Z-value)(−2.584)***(−1.569)(−2.013)**
Herfindahl index (H5)     
Strong groupMean0.0050.0060.009
 Median0.0000.0000.003
Weak groupMean0.0160.0170.020
 Median0.0000.0030.008
DifferenceMean−0.011−0.011−0.012
 (t-value)(−2.664)***(−2.508)***(−2.756)***
 Median0.000−0.003−0.005
 (Z-value)(−2.594)***(−1.568)(−2.050)**
Panel B: Broad Definition
   IPO IPO Year End 2007
Herfindahl index (H3)     
Strong groupMean0.0050.0060.009
 Median0.0000.0000.004
Weak groupMean0.0160.0160.019
 Median0.0000.0040.008
DifferenceMean−0.011−0.010−0.010
 (t-value)(−2.732)***(−2.838)***(−3.088)***
 Median0.000−0.004−0.004
 (Z-value)(−2.965)***(−2.218)**(−2.540)**
Herfindahl index (H5)     
Strong groupMean0.0050.0070.009
 Median0.0000.0000.005
Weak groupMean0.0160.0170.020
 Median0.0000.0030.008
DifferenceMean−0.011−0.010−0.011
 (t-value)(−2.770)***(−2.862)***(−3.249)***
 Median0.000−0.003−0.003
 (Z-value)(−2.975)***(−2.242)**(−2.553)**

The concentration of outside blockholdings is always lower for firms in the strong group than for firms in the weak group. When classified by the narrow definition, at the time of IPO, the medians of H3 and H5 are zero for firms in both groups. However, at the end of the IPO year, the medians of H3 and H5 for firms in the strong group are still zero, while the medians of H3 and H5 for the firms in the weak group increase to 0.003. Thus, shortly after IPO, the concentration of blockholdings changes very little for firms in the strong group, while it increases for firms in the weak group. The results are even clearer when we use the broad definition. These results provide weak evidence that outside blockholdings are more diffused after an IPO for firms in the strong group than for firms in the weak group, consistent with our expectation. At the end of 2007, the differences in H3 and H5 between the two groups were even larger, having increased from 0.003 to 0.005 under the narrow definition, which is also consistent with our expectation that firms in the strong group are more likely to avoid outside blockholdings than are those in the weak group.

Stoughton and Zechner (1998) demonstrate that by using IPO underpricing, firms encourage large institutional investors to buy more shares and become blockholders who can monitor the firm and reduce agency costs. Based on their argument, firms in the strong group will have more concentrated ownership (excluding the controlling owner), because they have larger agency costs. However, this prediction is opposite to our prediction and our findings.

(v) Percentage of Shares Held by the Controlling Owner after an IPO

If our conjecture is correct that firms in the strong group are more likely to enhance control than are firms in the weak group, we expect less ownership reduction for firms in the strong group than for firms in the weak group. In this section, we investigate the percentage of shares held by the controlling owner at the time of an IPO, at the first year end after an IPO, and at the end of 2007, when the sample firms have been publicly listed for one to five years. The results are presented in Table 10.

Table 10. 
Percentage of Shares Held by the Controlling Owner/Family
Panel A: Narrow Definition
   IPO First Year After IPO 2007
  1. Notes: This table presents the percentage of shares held by the controlling owner/family at the time of an IPO, at the end of the IPO year, and at the end of 2007. Strong group includes firms with two generations of the family or siblings and/or cousins involved in the business. Weak group includes firms in which only the founding family member (or non-related founders) and their spouses are involved in the business. The narrow definition refers to family involvement defined based on the family relationship among board members while the broad definition refers to family involvement defined based on the family relationship among board members and senior managers. The mean difference is tested by the t-test while the median difference is tested by the Wilcoxon rank sum test. *significant at the 10% level; **significant at the 5% level; ***significant at the 1% level.

Strong groupMean65.0364.2057.35
 Median70.0068.6558.36
Weak groupMean61.8361.2552.29
 Median67.1066.6753.22
DifferenceMean3.202.954.13
 (t-value)(1.383)(1.256)(1.941)*
 Median2.901.985.14
 (Z-value)(1.552)(1.365)(2.026)**
Panel B: Broad Definition
   IPO First Year After IPO 2007
Strong groupMean65.4764.7357.68
 Median70.0069.0058.39
Weak groupMean61.1960.5651.57
 Median65.9365.0650.91
DifferenceMean4.284.176.11
 (t-value)(1.897)*(1.820)*(2.406)***
 Median4.073.947.48
 (Z-value)(1.955)*(1.982)**(2.407)***

The controlling owners in the strong group always have more ownership than the controlling owners in the weak group, whether measured by mean or median. When classified by the narrow definition, the controlling owners in the strong group hold an average of 65.03% (70% in median) of the shares, while the controlling owners in the weak group hold an average of 61.83% (67.10%) of the shares at the time of an IPO. There is little change at the end of the first year after an IPO, but by the end of 2007 the controlling owners’ average shares decline from 70% to 58.36% for firms in the strong group, and from 67.10% to 53.22% for firms in the weak group. The difference between the two groups increases from 2.90% at the time of the IPO to 5.14% at the end of 2007. The results are the same when the broad definition is used. Together, these results show that firms in the strong group tend to reduce ownership more slowly than do firms in the weak group, which is consistent with our expectation.

(vi) Sensitivity Test: After-IPO Performance

Extensive underpricing of firms in the strong group may indicate that investors expect poor performance as a result of family involvement after the IPO. A comparison of the one-, two- and three-year performance and industry-adjusted performances (ROE and cumulative abnormal return) after IPOs for firms in the strong group and firms in the weak group does not reveal significant differences in these accounting-based performance and market-based performance measures, as presented in Table 11. This result indicates that the difference in IPO underpricing is not due to after-IPO performance.

Table 11. 
Performance Comparison after IPO: Strong versus Weak
Panel A: Accounting Performance
Narrow Definition Mean Median
Weak Group Strong Group P-value Weak Group Strong Group P-value
  1. Notes:This table presents the mean and median comparison of firm performance after an IPO between firms in the strong group and firms in the weak group. Strong group includes firms with two generations of the family or siblings and/or cousins involved in the business. Weak group includes firms in which only the founding family member (or non-related founders) and their spouses are involved in the business. The narrow definition refers to family involvement defined based on the family relationship among board members while the broad definition refers to family involvement defined based on the family relationship among board members and senior managers. ROEt+1 is measured as EBIT divided by shareholder's equity after one year of IPO. ROEt+2 is the average two-year ROE after an IPO. ROEt+3 is the average three-year ROE after an IPO. CARt+1, CARt+2, and CARt+3 are the buy-and-hold abnormal return for 12-months, 24-months and 36-months after an IPO, respectively. *significant at the 10% level; **significant at the 5% level; ***significant at the 1% level.

Raw Performance       
ROEt+10.2340.2290.7380.2170.2180.945
ROEt+20.2270.2170.7180.2240.1980.303
ROEt+30.2300.2040.3250.2300.2060.412
Industry Adjusted Performance
ROEt+1−0.035−0.0100.196−0.0330.0060.264
ROEt+2−0.035−0.0240.563−0.014−0.0240.537
ROEt+3−0.027−0.0400.688−0.006−0.0330.412
Broad Definition Mean Median
Weak Group Strong Group P-value Weak Group Strong Group P-value
Raw Performance       
ROEt+10.2320.2370.7100.2110.2190.678
ROEt+20.2220.2200.9110.2240.2030.417
ROEt+30.2260.2050.3240.2300.2060.421
Industry Adjusted Performance
ROEt+1−0.050−0.0050.093*−0.0330.0060.148
ROEt+2−0.041−0.0180.284−0.015−0.0190.685
ROEt+3−0.032−0.0360.873−0.012−0.0290.788
Panel B: Market Performance
Panel A: Accounting Performance
Narrow Definition Mean Median
Weak Group Strong Group P-value Weak Group Strong Group P-value
CARt+10.1780.0720.401−0.052−0.0280.990
CARt+20.198−0.1200.041**−0.170−0.2150.701
CARt+30.1200.0330.647−0.292−0.3150.892
Industry Adjusted Performance
CARt+10.027−0.0580.308000.978
CARt+20.030−0.0620.424000.557
CARt+30.004−0.0080.919000.643
Broad Definition Mean Median
Weak Group Strong Group P-value Weak Group Strong Group P-value
Raw Performance       
CARt+10.1800.0830.426−0.0452−0.0280.990
CARt+20.183−0.0530.135−0.193−0.2040.946
CARt+30.1290.0300.595−0.292−0.3150.904
Industry Adjusted Performance
CARt+10.031−0.0530.301000.605
CARt+20.013−0.0210.758000.500
CARt+30.014−0.0240.747000.904

(vii) Sensitivity Test: Alternative Means to Retain Control

Besides using IPO underpricing to diversify shareholder base, there are other methods for retaining control. One way is to set up a family trust,8 which could restrict ownership selling and prevent family ownership dilution. For those firms, IPO underpricing may not be an important strategy to retain control. To investigate this issue, we first exclude firms with a family trust as the controlling owner and re-test the equation (1). The results are presented in Panel A, Table 12.

Table 12. 
Regression Estimates of the Relationship between IPO Underpricing and Family Involvement: the Effect of Family Trusts
Panel A: Excluding Firms Controlled by Family Trusts
Variable Narrow Definition Broad Definition
(1) FDR (2) Adj. FDR (3) FDR (4) Adj. FDR
  1. Notes: This table presents the regression results estimating the relation between IPO underpricing and issuers’ characteristics, taking the family trusts into consideration. The dependent variable is IPO underpricing measured by FDR and Adj. FDR. Family is a dummy variable equal to 1 if the firm is in the strong group, and 0 if the firm is in the weak group. Strong group includes firms with two generations of the family or siblings and/or cousins involved in the business. Weak group includes firms in which only the founding family member (or non-related founders) and their spouses are involved in the business. The narrow definition refers to family involvement defined based on the family relationship among board members while the broad definition refers to family involvement defined based on the family relationship among board members and senior managers. Trust is a dummy variable equal to 1 if the firm's controlling owner is a family trust and 0 otherwise. Retained is the percentage of ownership retained by substantial shareholders. Sold is the percentage of ownership sold by insider owners in the IPO process. Commitment is a dummy variable equal to 1 if the firms use firm-commitment offering, and zero if the firms use best-effort offering. Fixedprice is a dummy variable that is equal to 1 if the issuers use the fixed pricing method, and zero if the issuers use the book-building method. Sponsor is a dummy variable equal to 1 if the sponsor or one of the sponsors is on the ranking list and zero otherwise. Auditor is a dummy variable equal to 1 if the financial statements are audited by a Big Four audit firm and zero otherwise. OS is the offering size, measured as the number of shares offered times the offer price. Age is the number of years from the year of incorporation to the date of the IPO. ROS is measured as net income divided by sales in the year prior to the IPO. Debt is measured as total debt divided by total assets in the year prior to the IPO. CV is the cash flow rights divided by the voting rights of the majority owner at the time of the IPO. The heteroskedasticity adjusted White-t statistics are in parentheses. *significant at the 10% level; **significant at the 5% level; ***significant at the 1% level.

Intercept−22.006−24.689−21.771−24.545
 (−0.618)(−0.700)(−0.622)(−0.708)
Family12.713**12.763**12.740***12.738***
 (2.485)(2.498)(2.653)(2.657)
Retained−16.281−16.458−17.239−17.391
 (−1.161)(−1.196)(−1.215)(−1.249)
Sold−10.998−11.035−10.542−10.575
 (−1.507)(−1.533)(−1.440)(−1.465)
Commitment13.828**14.142**12.857**13.173**
 (2.219)(2.269)(2.144)(2.198)
Fixedprice−0.674−0.408−1.115−0.836
 (−0.154)(−0.093)(−0.256)(−0.192)
Sponsor−6.780−6.672−6.497−6.377
 (−1.511)(−1.506)(−1.494)(−1.485)
Auditor−8.279*−8.179*−7.448−7.335
 (−1.707)(−1.728)(−1.546)(−1.551)
LogOS3.880**3.949**3.989***4.059***
 (2.555)(2.629)(2.703)(2.783)
Log(1+Age)−12.516**−12.601**−12.442**−12.513**
 (−2.140)(−2.147)(−2.145)(−2.150)
ROS−43.318***−44.129***−41.986***−42.766***
 (−3.073)(−3.127)(−3.088)(−3.146)
Debt−5.474−5.922−3.979−4.455
 (−0.491)(−0.541)(−0.358)(−0.409)
CV5.8025.8474.2444.286
 (0.724)(0.728)(0.527)(0.531)
Industry&year dummiesYesYesYesYes
No. of observations118118118118
Adj.R20.4880.4940.4900.496
Panel B: Including the Interaction between Trust and Family
Variable Narrow Definition Broad Definition
(1) FDR (2) Adj. FDR (3) FDR (4) Adj. FDR
Intercept−25.341−27.388−23.846−25.890
 (−0.846)(−0.931)(−0.817)(−0.903)
Family10.676**10.782**10.984***11.037***
 (2.397)(2.424)(2.700)(2.719)
Trust8.633*8.609*8.1608.229
 (1.733)(1.754)(1.482)(1.516)
Trust*Family−13.588**−13.825**−11.672*−12.040*
 (−2.013)(−2.073)(−1.769)(−1.847)
Retained−17.761−18.447*−18.475*−19.112*
 (−1.606)(−1.688)(−1.673)(−1.751)
Sold−8.557−8.829−7.727−7.995
 (−1.322)(−1.401)(−1.200)(−1.276)
Commitment12.088***12.092***11.531***11.512***
 (2.926)(2.953)(2.874)(2.904)
Fixedprice0.3210.493−0.443−0.272
 (0.089)(0.138)(−0.125)(−0.077)
Sponsor−5.959*−6.065*−6.117*−6.201*
 (−1.708)(−1.758)(−1.775)(−1.820)
Auditor−8.384**−8.233**−7.521*−7.366*
 (−2.007)(−2.018)(−1.838)(−1.839)
LogOS3.470***3.548***3.457***3.533***
 (2.671)(2.777)(2.774)(2.882)
Log(1+Age)−11.933**−11.949**−11.712***−11.722**
 (−2.618)(−2.615)(−2.623)(−2.621)
ROS−34.738***−35.299***−35.014***−35.528***
 (−3.658)(−3.754)(−3.694)(−3.789)
Debt−11.807−11.934−11.278−11.418
 (−1.228)(−1.260)(−1.182)(−1.215)
CV4.8435.1193.9294.187
 (0.730)(0.768)(0.599)(0.635)
Industry&year dummiesYesYesYesYes
No. of observations151151151151
Adj.R20.4590.4660.4630.469

Compared with the full sample results presented in Table 6, the coefficients of Family are larger in all the four regressions in Panel A, Table 12. For instance, under the narrow definition of family involvement, and with underpricing measured by market adjusted first-day return, the coefficient of Family is 12.763 in Table 12, much larger than the result of 7.4 in Table 6. It indicates that firms with strong family involvement have even stronger incentive to underprice in IPO without the setting of family trusts. Then we re-test the equation (1) including the interaction between Family and Trust. The coefficients of the interaction are significantly negative in all the four regressions, as presented in Panel B of Table 12, suggesting that controlling by family trusts reduces the level of IPO underpricing of firms with strong family involvement. The negative coefficients of the interaction indicate that setting up a family trust is also a means of retaining family control and, accordingly, firms controlled by family trusts have less incentive to retain control through IPO underpricing.

(viii) Alternative Explanation 9

Family firms are usually regarded as having more private benefits of control and, thus, a desire to keep controlling rights within the family (Johnson et al., 2000; and Morck et al., 2004). Benninga et al. (2005) theorize the going public decision of an entrepreneur who arbitrates between private benefits of control when a firm is private and the higher price of the shares that more diversified public investors are willing to pay. Their study indicates that even some private benefits of control when a firm is private would be lost after IPO. IPO subscribers require to buy the new offering at a discount because those firms retain the option to get private benefits, especially for strong-group firms. Meanwhile, the owners of family firms give up less private benefits of control after IPO, and thus can accept a lower offering price, which may lead to correspondingly larger IPO underpricing than firms with less or no incentive to retain controlling rights within the family. Thus, their study provides an alternative explanation to our findings that the association between family involvement and IPO underpricing could be due to discount required by IPO subscribers to compensate for future expropriation by the largest shareholder.

To illustrate whether the findings in our study are driven, primarily, by this alternative explanation, we incorporate the loss of private benefits of control in an IPO into the main regression. Because private benefits of control are chiefly from controlling ownership, we assume that families with more ownership have both ability and incentive to extract more private benefits. The loss of private benefits in an IPO is measured by the ownership decreasing of the largest owner immediately after an IPO. The mean (median) of ownership decreasing is 23% (25%).

The results are presented in Table 13. Compared with the main tests in Table 6, the sample size here is somewhat smaller because we could not obtain the pre-IPO ownership data for four firms. PB is the ownership decreasing of the largest owner, immediately after an IPO. The coefficients of PB are insignificant, but all are negative in the four regressions, consistent with the theory of Benninga et al. (2005) that less loss of private benefits is associated with larger IPO underpricing. Meanwhile, the coefficients of Family are still significantly positive, showing that after considering private benefits loss suggested in the theory of Benninga et al. (2005), our results for the IPO underpricing still exist.

Table 13. 
Regression Estimates of the Relationship between IPO Underpricing and Family Involvement: Including Private Benefits of Control
Variable Narrow Definition Broad Definition
(1) FDR (2) Adj. FDR (3) FDR (4) Adj. FDR
  1. Notes: This table presents the regression results estimating the relation between IPO underpricing and issuers’ characteristics, including private benefits of control. The dependent variable is IPO underpricing measured by FDR and Adj. FDR. Family is a dummy variable equal to 1 if the firm is in the strong group, and 0 if the firm is in the weak group. Strong group includes firms with two generations of the family or siblings and/or cousins involved in the business. Weak group includes firms in which only the founding family member (or non-related founders) and their spouses are involved in the business. The narrow definition refers to family involvement defined based on the family relationship among board members while the broad definition refers to family involvement defined based on the family relationship among board members and senior managers. PB is the ownership decreasing of the largest owner immediately after an IPO, which is used to proxy for private benefits loss. Retained is the percentage of ownership retained by substantial shareholders. Sold is the percentage of ownership sold by insider owners in the IPO process. Commitment is a dummy variable equal to 1 if the firms use firm-commitment offering, and zero if the firms use best-effort offering. Fixedprice is a dummy variable that is equal to 1 if the issuers use the fixed pricing method, and zero if the issuers use the book-building method. Sponsor is a dummy variable equal to 1 if the sponsor or one of the sponsors is on the ranking list and zero otherwise. Auditor is a dummy variable equal to 1 if the financial statements are audited by a Big Four audit firm and zero otherwise. OS is the offering size, measured as the number of shares offered times the offer price. Age is the number of years from the year of incorporation to the date of the IPO. ROS is measured as net income divided by sales in the year prior to the IPO. Debt is measured as total debt divided by total assets in the year prior to the IPO. CV is the cash flow rights divided by the voting rights of the majority owner at the time of the IPO. The heteroskedasticity adjusted White-t statistics are in parentheses. *significant at the 10% level; **significant at the 5% level; ***significant at the 1% level.

Intercept−18.919−21.153−21.911−24.315
 (−0.604)(−0.688)(−0.737)(−0.834)
Family7.782*7.840*8.648**8.610**
 (1.934)(1.948)(2.308)(2.297)
Retained−15.893−16.492−18.034−18.596*
 (−1.407)(−1.477)(−1.596)(−1.667)
Sold−7.080−7.304−6.964−7.184
 (−1.078)(−1.137)(−1.055)(−1.113)
Commitment10.910**11.053**10.980**11.109**
 (2.396)(2.444)(2.461)(2.510)
Fixedprice−0.654−0.450−0.689−0.471
 (−0.171)(−0.118)(−0.185)(−0.126)
Sponsor−5.676−5.795−5.938−6.039
 (−1.476)(−1.517)(−1.562)(−1.599)
Auditor−6.330−6.172−6.244−6.078
 (−1.434)(−1.433)(−1.435)(−1.430)
LogOS3.144**3.221**3.428***3.508***
 (2.393)(2.493)(2.768)(2.883)
Log(1+Age)−11.372**−11.384**−11.328**−11.315**
 (−2.345)(−2.340)(−2.403)(−2.392)
ROS−40.301***−40.801***−39.950***−40.374***
 (−3.573)(−3.643)(−3.739)(−3.811)
Debt−15.261−15.401−13.746−13.920
 (−1.477)(−1.509)(−1.320)(−1.354)
CV6.7487.0795.7566.085
 (1.017)(1.066)(0.861)(0.910)
PB−15.044−15.532−15.562−16.018
 (−0.794)(−0.818)(−0.810)(−0.833)
Industry&year dummiesYesYesYesYes
No. of observations147147147147
Adj.R20.4330.4390.4420.447

5. CONCLUSIONS

This paper examines difference in the level of IPO underpricing between Hong Kong firms with different levels of family involvement. We find a higher level of IPO underpricing in firms with strong family involvement than in firms with weak family involvement. We also examine the number of second- and third-generation family members, as well as siblings and cousins, involved in the business, and find a positive association between the number of such family members and the level of IPO underpricing, showing that firms with a greater potential for ownership diffusion among family members are more likely to underprice in order to avoid outside blockholdings. In addition, we find evidence that firms with strong family involvement attract more subscription for their shares, have less concentrated outside blockholdings, and reduce ownership more slowly than do firms with weak family involvement. The results are consistent with our prediction that, in order to retain managerial control, family owners use underpricing to reduce outside blockholdings. Their primary objective is to avoid monitoring from outside shareholders through oversubscription and rationing in the share-allocation process.

According to Benninga et al.'s (2005) theory on tradeoff between loss of private benefits of control and the gains of diversification, one concern about our paper is that the owners of family firms can accept a lower offering price, which may lead to correspondingly larger IPO underpricing, because owners of family firms give up less private benefits of control after IPO. To relieve this concern, we incorpor-ate private benefits loss that occurs during an IPO, measured by the ownership decreasing of the largest owner immediately after an IPO, in our main regression and our empirical results still hold. We also find that firms controlled by family trusts are less likely to retain control through IPO underpricing, suggesting that IPO underpricing and the family trust are effective substitute methods for retaining family control.

Our study suggests that the main purpose of a family firm's going public in markets with weak property rights protection is to seek financing, rather than to eventually sell the firm. Prior studies have found that wealthy families utilize pyramid structures to maintain control (e.g., Almeida and Wolfenzon, 2006; Johnson et al., 2000; and Morck et al., 2004) in environments with insufficient investor protection. Our study indicates that IPO underpricing may be just another way for family firms to retain control.

Footnotes

  • 1

    The full content is listed in http://www.hklii.org/hk/legis/en/ord/32/

  • 2

    Laven and Levine (2007) classify owners into five types: a family, a state, a widely held financial institution, a widely held corporation, and others.

  • 3

    Britain's Legacy Still Lives on in Hong Kong, Financial Times (22 February 2007).

  • 4

    Asymmetic information contained in listing by introduction is much less serious than that in the other types of listing and thus we exclude firms listed by introduction from our sample.

  • 5

    Rules Governing the Listing of Securities on the Stock Exchange of Hong Kong, Practice Note 18.

  • 6

    In an effort to remain concise, we do not tabulate the results with returns in five trading days in the paper, but they are available upon request from the authors.

  • 7

    We acknowledge that it is possible for control of firms in certain industries to be retained by the families when there is little professional knowledge required or when the business is highly relationship-based. Many factors may influence the distribution between firms in the strong group and those in the weak group. However, since the industry distribution is not the focus of our study, we control only for the industry effect in the regression models.

  • 8

    Another means for family firms to retain control is dual-class structure. However, the dual-class structure has been banned in Hong Kong since 1991.

  • 9

    We thank the referee for putting forward the alternative explanation to our findings.

Ancillary