Does Internal Capital Facilitate Access to External Financing? Evidence from IPOs by Family Business Groups *

Transcription

1 Does Internal Capital Facilitate Access to External Financing? Evidence from IPOs by Family Business Groups * Ronald W. Masulis Australian School of Business University of New South Wales, Sydney, NSW 2052, Australia Phone: , ron.masulis@unsw.edu.au Peter Kien Pham Australian School of Business University of New South Wales, Sydney, NSW 2052, Australia Phone: peter.pham@sydney.edu.au Jason Zein Australian School of Business University of New South Wales, Sydney, NSW 2052, Australia Phone: j.zein@unsw.edu.au May 15, 2014 * The authors would like to thank Heitor Almeida, Joseph Fan, Jonathan Karpoff, Amy Kwan, Michelle Lowry, Jan Mahrt-Smith, Vijay Marisetty, Randall Morck, Lilian Ng, Micah Officer, Jay Ritter, Weidong Xu, and conference participants at the 2010 Financial Management Association meeting, the 2011 Finance and Corporate Governance conference (La Trobe), the 2011 Frontiers in Finance conference (Alberta), the 2012 Financial Mangement Association Asian meeting, the 2013 NUS Family Firm conference, the 2013 Conference on International Corporate Governance & International Public Law, and the 2014 Finance Down Under conference, and seminar participants at the Australian National University, Chinese University of Hong Kong, Hong Kong University, La Trobe University, Monash University, Queensland University of Technology, University of South Australia, University of Technology Sydney, and University of Western Australia for helpful comments on different earlier versions of the paper. We are also very grateful to Alvin Ang, Yue Lu, and Stephanie Dash for their help with the data. All errors are our own.

2 Does Internal Capital Facilitate Access to External Financing? Evidence from IPOs by Family Business Groups Abstract Recent evidence highlights the benefits of family business group affiliation, but how they support affiliates is poorly understood. Studying groups in 44 countries, we show that group internal capital facilitates access to public equity markets for new firms. We find internal capital accumulation predicts which groups conduct IPOs, their timing, and the affiliates that sponsor and invest in new firms. Groups typically sell shares in larger IPO firms with difficultto-finance investments and in weaker capital market conditions. Group IPOs raise proportionally less external equity, incur lower flotation costs, and post-listing can better weather adverse capital market conditions than unaffiliated IPOs.

3 Across the world, the growth of new ventures is often severely constrained by their inability to raise external equity capital. In developed economies, this funding gap can be bridged by angel investors and venture capitalists. In most emerging economies, however, contracting mechanisms and property right protections are often insufficiently developed to support substantial venture capital activity. One alternative source of financing for young firms facing external capital constraints is equity capital investments by other corporations (Bena and Ortiz-Molina (2013)), in some cases structured as business groups, which are common in most countries and are especially dominant where equity markets are underdeveloped. While the extant literature has extensively analyzed the roles of venture capitalists in developed economies, much less is known about business group support for new firms and how their control structures can facilitate access to external capital markets. Prior studies of business groups suggest that they have both a dark side, arising from ownership structures that allow voting rights in excess of cash flow rights, which facilitate the extraction of private benefits of control (see La Porta, Lopez-de-Silanes, Shleifer and Vishny (2002) and Morck, Wolfenzon and Yeung (2005) for a review), and a bright side, stemming from group financing advantages associated with their reputation and access to internal capital (see Khanna and Palepu (2000), Gomes (2000), Almeida and Wolfenzon (2006) among others). For young firms with high investment needs, recent evidence indicates that the aggregate effect of group affiliation can be positive (Almeida, Wolfenzon, Park and Subrahmanyam (2011), Masulis, Pham and Zein (2011), and Bena and Ortiz-Molina (2013)). However, the specific channels which provide financing advantages to member firms remain largely unexplored. One exception is Gopalan, Nanda and Seru (2007, 2013), who show that intra-group loans and dividends can be used to transfer internal capital within groups. Yet, to 1

4 date there is no evidence on how group financing advantages directly support their member firms in the process of raising external funding. Our study focuses on initial public offerings (IPOs) of group-affiliated firms to investigate how a group s internal capital can facilitate access to the public equity market for its young affiliated firms by bridging the capital gap associated with accessing the external equity market. In general, a young firm s access to new equity can be constrained by the price discount required to compensate external investors for the associated adverse selection and moral hazard risks, which can excessively dilute an entrepreneur s shareholding, prevent the firm from raising sufficient capital, and even delay profitable investment projects. Existing theories posit that in dealing with such external financing constraints, families and individuals who control groups possess certain financing advantages over independent entrepreneurs. Almeida and Wolfenzon (2006), for example, argue that when there is a need to set up a new firm to obtain external funding for projects with low pledgeable cash flows and high capital requirements, the controlling shareholder can utilize a pyramid structure to transfer internal capital from an existing affiliate (parent) to help meet some of the new firm s financing needs. By doing so the group minimizes the amount of external capital raised and its associated costs. Without such internal capital support, unaffiliated firms with similar characteristics face a serious disadvantage in raising external capital. In another model, Gomes (2000) arrives at a similar prediction regarding how external financing constraints can be overcome in a weak governance environment: a controlling family s retained ownership percentage in an IPO firm serves as a positive signal and an effective bonding mechanism to commit the family not to expropriate minority shareholders. Without large shareholdings by IPO insiders, it will be difficult for such firms to go public. 2

5 This highlights a pyramid s financing advantage, as the ability to raise external capital through successively creating new firms, rather than selling stakes in existing firms, allows a controlling family to maintain highly concentrated ownership positions in its affiliated firms. Some evidence consistent with these theoretical explanations is documented in Almeida et al. (2011), Masulis, Pham, and Zein (2011), and Bena and Ortiz-Molina (2013). They show that firms with difficult to finance characteristics are selected into the bottom of pyramidal structures. In this paper, rather than relying on firm placement decisions within a group as evidence in favor of the operation of an internal capital support channel, we seek to examine the operation of this channel more directly by analyzing the actual transactions through which a group s advantages in raising external capital should clearly emerge, namely an initial public offering. Focusing on these transactions provides an empirical setting where the underlying internal capital support mechanism that groups can utilize to supplement their external capital raising efforts can be more directly observed and tested. Further, by studying IPO events we can more accurately trace family group characteristics that dictate their fundraising choices and pyramiding decisions. Our setting also allows us to directly compare external fundraising outcomes between groups and similar unaffiliated listings. In doing so we document a number of fundraising advantages of going public group affiliates arising from group support, relative to other IPO firms and thereby provide new empirical insights that explain why family groups are able to expand the number of publicly listed firms under their control. Our IPO sample includes 12,683 firms from 44 countries going public during the period. We identify whether an IPO firm is affiliated with a family-controlled business group, and for comparison purposes, those affiliated with non-family business groups 3

6 (controlled by widely held firms, financial institutions or governments). Distinguishing these two types of business groups is important because the internal capital support channel discussed in Almeida and Wolfenzon (2006) is likely to be less relevant for non-family groups because there is no single controlling shareholder with the authoritative power or incentives to internalize the long term benefits of within-group capital allocation decisions. If an IPO firm is not affiliated with any group, we identify whether a pre-ipo controlling owner exists and whether it is an individual/family, government, widely held corporation, or financial institution. To our knowledge, this is the first cross-country dataset on pre-ipo owner identities. Using this data, we show that IPOs by family and non-family groups represent an important slice of the global new issues market, making up an average of 22% of IPO issuers and 29% of the aggregate IPO proceeds across our sample of 44 countries. Our empirical analysis focuses on the most direct implication of the theory of group financing advantages for IPO transactions: that family groups utilize internal capital to partially meet the financing requirements of large, difficult-to-finance projects that cannot be funded wholly by a group s internal capital, so that they can raise external funding as a new firm. If an IPO is indicative of a need to pursue sizable investment opportunities then we should observe that family business groups with the greatest accumulation of internal capital should be the most capable of taking new affiliates public. To examine this prediction, we construct yearly snapshots of the structure and membership of all business groups for our sample period, regardless of whether they tap the IPO market, and analyze group characteristics leading up to an IPO. Our main focus is on variations in internal capital accumulation, measured by additions to retained earnings at both the group level and the group member level. We find that within the cross-section of family business groups, the rate of group-level 4

7 internal capital accumulation significantly predicts which groups conduct an IPO, even after controlling for group aggregate growth opportunities and investment requirements. Analyzing individual groups across time, an IPO event is more likely to occur after a large increase in a group s internal capital accumulation. This relationship does not appear to be driven by a going public affiliate s own profitability, as it continues to hold after we exclude such affiliates contributions to the retained earnings of the sponsoring groups. Within each individual family group, we document that additions to retained earnings at the firm level help explain which existing affiliates (1) become IPO parents by bringing a subsidiary public (as well as the timing of such transactions), and (2) provides additional intragroup funding by other affiliates share purchases of minority stakes in affiliated IPOs. Taken together, these findings are consistent with the hypothesized role that group internal capital plays in facilitating access to external equity capital, particularly for first time equity capital raisings through an IPO. We check for whether the above evidence simply reflects the possibility that group firms tend to time their external capital raisings around strong performance periods. Delving into sources of group firm internal capital accumulation, we find that group IPO events are more likely to occur not only when strong operating performance creates large additions to retained earnings, but also when group members deliberately retain internal capital by limiting their dividends. Thus, taking an affiliate public does not appear simply to be a strategic decision to take advantage of positive earnings shocks. We further address this issue by examining the choice family groups make between raising capital through an IPO of an unlisted affiliate firm or through a seasoned equity offering (SEO) by an existing listed affiliate. If our results are driven by groups opportunistically raising capital to take advantage of positive earnings shocks, then we should observe that additions to retained earnings are 5

8 equally important in explaining SEO decisions. Comparing the choice between an IPO and an SEO, we find that group internal capital accumulation is only relevant for IPOs, consistent with a hypothesized group financing advantage. It is important to note that by documenting a positive relationship between group internal capital accumulation and an IPO s likelihood, we do not dismiss the possibility that group internal capital can also be employed to prolong the funding of certain investment projects as private entities within a group. However, it is difficult to directly analyze this alternative funding approach due to data limitations associated with the unobservable nature of the characteristics of group firm divisions and private firms. Nevertheless, we are able to indirectly gauge the effect of such group financing support of affiliates during their development as an unlisted entity by analyzing the offer size decisions of IPOs in our sample. The group financing advantages prediction suggests that at least for ventures that eventually require external financing through an IPO, the roles of prior internal investments should be manifested in the ability of the firm to go public by raising a relatively smaller amount of new capital compared to unaffiliated firms, despite having larger investment needs. Consistent with this prediction, we observe that family group IPO firms on average raise relatively smaller proportion of new equity than do non-group IPO firms. Within the former cohort, the proportion of new equity raised is strongly related to the retained earnings accumulations of both the sponsoring group in aggregate and especially to its parent firm. In addition to documenting the characteristics of groups and group firms that sponsor IPOs, we examine the value of having an internal capital support channel by exploring differences between group IPOs and non-affiliated IPOs. If group internal capital provides significant funding advantages in the IPO process, groups should be less sensitive to unfavorable market conditions that normally make issuing external capital relatively more 6

9 expensive and success more uncertain. Thus, empirically we should observe that they are able to tap the IPO market under a wider range of capital market conditions. This same advantage should also allow them to float stock in firms which would ordinarily find it difficult to raise external equity, and to do so at lower flotation costs than other non-group firms. In the long run, group IPO firms should also exhibit a superior ability to maintain continued access to public equity market thereafter. We investigate the differences between group and non-group IPOs along these dimensions and in general, document evidence broadly consistent with a positive role of group affiliation. First, we find that the relative frequency of family group IPOs is greater in countries with stronger external financing constraints. Within a country, the proportion of family group firms going public appears to be less cyclical with respect to the frequency of new listings, indicating that family groups can tap public equity markets for their affiliates under less favorable capital market conditions. This is an important financing advantage since it allows group firms to raise external equity when they have profitable investment opportunities, rather than being forced to wait for a more favorable phase of the IPO cycle. Second, we document that relative to independent IPO firms, group affiliates going public (especially those controlled through a pyramid) tend to be larger, but are less established firms, and in fact, have more intangible assets, lower pledgeable cash flows, and higher capital expenditure needs. In other words, these firms appear to have large funding needs and high asymmetric information, making them the types of firms that face the greatest difficulties in going public. Family groups are twice as likely to place their new IPO firms at the bottom of a pyramid compared to directly owned family firms. Given their characteristics, IPO firms stand to realize substantially greater benefits from a group s internal capital support than would a typical unaffiliated IPO firm especially under a pyramid structure. 7

10 Third, we examine differences in underpricing and underwriter fees across IPO firms to test whether group support translates into an ability of affiliates to go public at lower cost. Compared to independent IPO firms, we find group-affiliated issuers, especially those structured as a part of a pyramid, have significantly lower underpricing and lower overall flotation costs (incorporating both underpricing and underwriting fees). These findings do not arise simply because group affiliated IPO firms are essentially carve-outs, which could have lower flotation costs because as divisions of an already listed firm, they are relatively more visible and established than unaffiliated IPO firms. In particular, we show that IPO firms carved out of widely held non-group firms or even privatizations of government-owned firms do not realize the same low flotation costs. Further, IPOs controlled through dual-class shares experience higher underpricing than other types of IPOs. This is consistent with Almeida and Wolfenzon s (2006) argument that pyramidal groups provide a unique financing advantage, not offered by other control enhancing mechanisms such as dual-class shares. Finally, we analyze how financing advantages of group IPOs manifest themselves in the post-listing period. On average, family group IPOs using a pyramid structure appear to be less capital constrained, indicated by their lower SEO and capital expenditures sensitivity to external capital market conditions than either all other non-group IPO firms or IPO firms with dual-class shares. This suggests that from the perspective of an IPO as the start of an extended external capital raising process, initial (and potentially ongoing) internal capital support by groups also appears to alleviate significant financing frictions faced by young firms in the aftermaket, an advantage which pure control enhancing mechanisms such as dual-class shares fail to provide. We also attempt to document the value of such support by examining differences in relative market valuation between group and non-group IPO firms five years after listing controlling for a potential endogenous selection problem that observed 8

11 independent listed firms on average have superior qualities that allow them to go public without relying on group support. Using a treatment effects model with IPO market conditions around listing dates as instruments, we find evidence to indicate that group affiliation enhances an IPO firm s valuation relative to the counter-factual situation in which the same firm raises funds as an independent firm. Our evidence has important implications for the ongoing debate on the raison d'être for family business groups. Various studies point out that a deviation of cash flow rights and control rights, which is an inherent characteristic of pyramidal business groups, can increase the risk of minority shareholder expropriation (Bae, Kang, and Kim (2002), Claessens, Djankov, Fan and Lang (2002), La Porta et al. (2002), Lemmon and Lins (2003)). This evidence has raised a puzzling question regarding why rational minority investors would continue to co-invest alongside controlling shareholders. Our study adds to the body of research seeking to reconcile the risk of business groups and their prevalence by documenting an important external financing channel through which such business structures benefit both their members and minority shareholders: namely, by helping difficult-to-finance firms access the IPO market. Our large international sample of IPOs also helps provide complementary, but more direct transaction-based evidence regarding the roles of pyramids in alleviating financing constraints in high-growth affiliated firms (Masulis et al. (2011), Bena and Ortiz-Molina (2013)). In contrast to Bena and Ortiz-Molina (2013), who examine European privately held firms, IPOs of listed groups allow a pyramid s financing advantages to be magnified since public investor participation creates an empirical setting with much greater financing frictions. From the vantage point of a sample of newly listed firms, the weight of evidence we uncover fails to support expropriation motives as the primary reason for adding new firms to family-controlled pyramids. This suggests that on average, the private benefits of control in 9

12 young, high-growth firms are small and that more efficient structures exist for maintaining corporate control, such as dual-class shares. For these firms, our results consistently point to significant financing advantages of group affiliation. However, we acknowledge that with respect to our findings on the post-listing outcomes of group affiliation, we cannot dismiss the existence of other group benefits such as providing reputation capital and implicit guarantees (Khanna and Palepu (2000), Khanna and Yafeh (2005), Gopalan, Nanda and Seru (2007)). Our study also contributes to improving our limited understandings of business group involvement in international IPO markets. Dewenter, Novaes and Pettway (2001) and Marisetty and Subrahmanyam (2010) have examined IPOs of business groups, but focused only on IPO underpricing within a single country. Other studies have examined cross-country variations in IPO underpricing (such as Boulton, Smart and Zutter (2010)), but not the role group structures play in shaping the many facets of IPO process. Another emerging area of research to which our findings are relevant is the financing of innovation. Belenzon and Berkovitz (2010) find that business groups can spur innovation in industries where access to external capital is typically poor. Our study further clarifies this channel by showing that to the extent that IPO firms are investing in innovation-intensive projects, family groups can help create access to public capital to support the growth of these ventures, particularly in markets where financing innovation is challenging. Our paper proceeds as follows. Section I describes the IPO and group samples. Section II examines group characteristics that facilitate an IPO event. Section III documents variations in relative group IPO frequency across market conditions. Section IV investigates differences in firm characteristics and flotation costs between group and non-group IPOs. Section V analyzes the post-listing impact of group affiliation. Section VI concludes. I. Data and Sample Selection 10

13 Our initial sample consists of international initial public offerings obtained from the Thomson Reuters SDC Platinum New Issues database during the January 1997 to December 2007 period. To maximize our data coverage, we use another source, the Zephyr database from Bureau Van Dijk, to gather data on other IPOs not covered by SDC Platinum. From this combined dataset, we exclude IPOs by investment funds, partnerships and trusts due to their distinct structures, investment objectives, and regulatory constraints. We further eliminate issues that do not involve common stock and firms that raise less than US$0.5 million, which are likely to be back-door listings. The latter requirement also means that our analysis excludes spin-offs, split-offs and demergers where the firm raises no new capital. For each IPO, we extract from SDC Platinum or Zephyr key filing information, including parent firm identity (if available), issue amount, offer price, underwriter identity, fee information, number of shares offered, and an indicator for venture capital backing. These are supplemented by data on post-listing share prices and total number of outstanding shares from the Thomson Reuters Datastream and Worldscope databases. We then apply a comprehensive identification procedure to classify the type of preissue ownership structures of our sample IPO firms, that is, whether an IPO firm is affiliated with a business group, and if not, whether it has a controlling shareholder. Our starting reference point is the business group database constructed by Masulis et al. (2011), which contains a snapshot of ownership information for 28,635 international firms from 45 countries as of Masulis et al. (2011) define a group as two or more listed firms in the same market, linked together by one common controlling shareholder, where control is defined as holding a minimum of 20 percent of voting rights, or 10 percent if the shareholder also holds other forms of control such as being a founder, CEO or chairman of the board. Based on this 11

14 data, IPOs occurring prior to or during 2002 are matched to member firms of known business groups in For each matched IPO firm, we cross check with other data sources (SDC Platinum, Factiva, company websites and Google searches) to ensure a matched firm does not become part of a group through a post-ipo acquisition. Firm delisted before 2002 and those listed after 2002 do not appear in the Masulis et al. (2011) business group database, and therefore to assess whether they are group-affiliated, we examine the ownership structures of their parents. Parent information is available for some of the IPOs in SDC Platinum (or Zephyr). For IPOs whose parent company is not listed in SDC Platinum (or Zephyr), we collect their ownership structures for the year immediately after their listing date from the Thomson Reuters Worldscope database and the Bureau van Dijk Orbis database. This is supplemented with several other data sources that provide annual ownership information, including some stock exchange / securities regulator websites (in Belgium, India, and Italy), the Taiwan Economic Journal database of East Asian companies, and company annual reports (obtained from the Mergent Online database and the Thomson Reuters OneSource database). We then assume that an IPO firm has the same controlling shareholder/parent at the listing date (with control defined in the same way as in Masulis et al. (2011)) as it does in the financial year immediately after listing. For IPO firms with no immediate post-listing ownership information, we identify their pre-ipo owners by manually searching through prospectus filings, pre-quotation announcements from stock exchanges / securities regulators, and financial news articles around the time of listing. 2 If the parent is a corporation, we attempt to match it with groupaffiliated firms identified in the Masulis et al. (2011) sample, while if it is not part of a known 2 Pre-quotation announcements and financial news articles related to an IPO are primarily obtained from Factiva. Prospectuses are obtained from stock exchange and securities regulator websites. 12

15 group based on the 2002 group data, we employ the same group identification procedures used in Masulis et al. (2011) to determine whether it is listed and if so, who is its ultimate owner. This allows us to identify and construct a substantial number of new groups that are not in the Masulis et al. (2011) database. If an IPO firm is identified to be group-affiliated, the firm is further classified according to its ultimate owner: family groups (those controlled by an individual entrepreneur or a family) and non-family groups (those controlled by a widely-held company, financial institution or government). This distinction is made because the incentives and degree of centralized control can differ across these ultimate owner types. If an IPO firm is not groupaffiliated, we also classify it by whether it is controlled by a family (an individual entrepreneur or group of founders), a government body, a widely held corporation or a financial institution. To investigate the characteristics of each business group immediately before a member firm undertakes an IPO, we trace the evolution of each family (and non-family) business group in terms of their affiliated firms from 1997 to 2007 in order to obtain an annual snapshot of each group over our sample period. This task is aided by using information from our IPO database, which identify additions to each group as well as information on other transactions that connect a new firm or disassociates an existing member from the group. Specifically, we obtain mergers and acquisitions data from SDC Platinum to identify group firms that are either acquired into or divested from a group, and de-listings data from Datastream to identify firms no longer in a group for other reasons (such as bankruptcy and going private). As a final verification, our group data are cross-checked against a snapshot of firm ownership information taken at the end of our sample period from the Thomson Reuters Worldscope database and the Bureau van Dijk Orbis database to pick out any remaining group firms not 13

16 identified by the above steps. This procedure allows us to assemble a yearly snapshot of all business groups in our dataset, even those that do not conduct IPOs during the sample period. Using the data described above, we identify the organizational structure which directly facilitates the going public of a family group firm. There are two possible ways a family group can form or expand by listing a new firm. The first is where a public firm controlled by a family lists a subsidiary, in a carve-out or a partial spin-off (note that we do not include the latter case if there is no new capital raised), creating or expanding a pyramid. The second method by which a group s organizational structure can evolve is horizontally, defined as when a family takes public a privately held firm currently under its control, and retains a substantial direct ownership position. Similar to the Masulis et al. (2011) database, we can identify the position of each member firm in the business group organization, captured by the number of pyramid layers (other listed firms in the control chain) separating a firm from its controlling shareholder. Our final sample consists of 12,683 IPOs from 44 countries, for which the identity of the pre-ipo controlling shareholder (or whether the firm is widely held) can be ascertained. Our ownership identification procedure is quite exhaustive as these 12,683 IPOs make up on average about 96% of the IPOs in each country that satisfy our sample selection criteria. Table I provides details on this coverage for each country and also reports the cross-country differences in the number of IPOs and total gross proceeds of family group IPOs, non-family group IPOs and those going public without group backing, classified according to the type of controlling shareholder. Overall, group affiliated IPOs are an important segment of IPO markets around the world, on average representing about 22% of IPO issues (16% for family 14

17 group firms and 6% for non-family group firms) and 29% of aggregate IPO proceeds (17% for family group firms and 13% for non-family group firms) in each country. [INSERT TABLE I HERE] II. Group-level Analysis In this section, we examine family business group characteristics that influence their decision to bring public an affiliate or division of a listed member to the IPO market. Consistent with the theory in Almeida and Wolfenzon (2006), we hypothesize that one channel through which groups (and especially pyramids) provide financing benefits to member IPO firms is through accumulated retained earnings within a group. These group earnings represent a source of equity capital for difficult-to-finance member firms planning to go public to bridge their unmet external funding needs. If an IPO is generally considered an important milestone that opens up on-going access to large public sources of capital, we should observe that family business groups with strong internal capital actively exploit their financing advantages in the IPO market. This theoretical framework provides several empirical predictions related to the likelihood of observing a group IPO event. Across different groups, we expect those that retain a larger fraction of their earnings to be more active in the IPO market, and the timing of group IPOs to be related to the timing of a group s earnings and its payout decisions. Within a group, we should observe evidence indicative of an internal capital market in operation with respect to IPO activities. The affiliate selected to float a subsidiary as a new firm should have higher retained earnings than other affiliates. In addition, intragroup investments are also typical of family business groups. In the case of a 15

18 new IPO, these investments should come from existing affiliates with superior internal resources. A. Aggregate Group Characteristics That Influence Group IPO Decision In order to test the above prediction, we utilize information on all family (and nonfamily) groups and not just those that conduct IPOs, as well as their structures on a yearly basis over the period. We examine aggregate financial information for every group in each financial year, excluding affiliated firm(s) going public in the same year. The main variable of interest in our analysis is the rate of internal capital accumulation. This is measured by the aggregate yearly additions to a group s retained cash earnings, separated into two components. GRPNETINCOME is calculated as the group s total net cash income (net income excluding depreciation, amortization, and impairment charges) and GRPDIVIDEND is its total cash distributions, both scaled by the group s year-beginning total assets. 3 This separation allows us to assess whether profit accumulation and cash distribution, which have opposite impacts on the availability of internal capital, play different roles in predicting which groups sponsor IPOs. It is important to note that although our reported analysis is based on a yearly construction of these variables as it suits the panel structure of our data, we also examine 3- year averages of GRPNETINCOME and GRPDIVIDEND in several tests that do not involve strictly year-to-year variations. This alternative construction captures a more long-term buildup in internal capital, yet we find qualitatively the same results in these tests. We consider a number of other group-level characteristics that can explain their ability to expand group assets by raising external equity capital for new public firms. In particular, we capture the potential effect of group size using two variables: the number of member firms in 3 Our results are qualitatively similar if we use calculate these variables not at the aggregate group level but using the median across group firms. 16

19 each group (GRPNOFIRMS) and the average size of member firms in each group (GRPFIRMSIZE). To describe a group s current control structure, we use an indicator variable for whether dual-class shares are currently in use in the group (GRPDUALCLASS) and the average pyramid layer value of all existing affiliates (GRPPYRAMID). 4 We also control for several important financial indicators, including (1) the aggregate level of debt (GRPDEBT), as this can be an alternative source of external financing to fund pre-ipo expansion, (2) the value weighted average Tobin s Q for each group (GRPQ), as this indicates future growth opportunities, and (3) group-wide investment activity (GRPCAPEX), as this captures immediate investment requirements. A group s industry diversity can influence its incentive to create new listed firms and our analysis accounts for this possibility using an indicator variable for the existence of a financial firm member (GRPFINMEMBER) and a Herfindahl index of industry diversification within a group based on the primary SIC codes of its existing members (HERFINDAHL). We further hypothesize that more established groups have a reputational advantage that helps with the process of raising external capital. Our analysis thus includes group age, constructed as the natural logarithm of the age at listing (in years) of a group s oldest firm (GRPAGE). Existing relationships with a reputable investment banker may have a similar effect, which is captured by an indicator variable for when a group member has hired a top-quintile underwriter for another IPO or SEO in the prior 3 years (GRPPRETOPUW). The last set of control variables are related to IPO market conditions in a particular year. For a given country-year, INDEXRET and IPOVOL are respectively the country s MSCI index return and the number of successful IPOs in the year scaled by the total number of IPOs in the entire sample period for the same 4 This value is based on the number of other listed firms in the chain of control between a group firm and the ultimate controlling shareholder (directly held firms have a pyramid value of zero) 17

20 country. In Table II, we report on tests of whether these variables significantly influence the likelihood of observing a group IPO event using different logit model specifications. In Column 1, we estimate a panel data random effects logit model, where for each group-year, the dependent variable is equal to one if the IPO listing date is within six months of a group s fiscal year-end date. Unobserved group effects, assumed for now to be uncorrelated with the errors, are the random effects in the model. In Column 2, we estimate a panel data random effects multinomial logit model that explicitly accounts for SEOs as the alternative external equity fund raising method, using the choice of either raising new equity through an IPO or an SEO as the dependent variable. 5,6 This model allows us to more effectively contrast the role of internal capital in supporting the going public process of a new group firm, relative to other situations where external equity capital is required, but does not require the incorporation of new firms. With both models, we use year fixed effects to facilitate a cleaner comparisons across groups. We find strongly supportive evidence of the proposition that a group s internal capital helps facilitate the entry of its affiliates into the equity capital market through an IPO, with both measures of additions to group retained earnings, GRPNETINCOME and GRPDIVIDEND, being significant. The signs of their coefficients are consistent with the internal capital support hypothesis. Additions to retained earnings from strong operating performance and by limiting dividend payouts, both serve to predict increases in the likelihood of a group sponsoring an IPO. In Column 3, our analysis attempts to explicitly control for unobservable group fixed effects by estimating a conditional logit model predicting which year each group conducts an 5 SEOs are also obtained from SDC Platinum based on the same selection criteria as applied to our IPO sample. 6 The estimation follow the method proposed by Rabe-Hesketh, Skrondal, and Pickles (2005) 18

21 IPO. This test therefore considers only those groups that experience an IPO event in our sample period. Focusing on within-group financial changes, we document that the conditions conducive to an IPO emerge when a group experiences improvements in operating performance and when it actively limits dividends in the year leading up to the IPO. This is consistent with groups timing their IPOs for periods when they can provide better internal capital support. The venture capital analogy is bridge financing where venture funds provide firms nearing an IPO with further capital to ensure they can successfully go public. It is possible that our results so far are driven by the need to strategically time any capital raising transaction to take advantage of a positive earnings shock by an existing group member. While this in itself would be a notable finding, our evidence on the negative relationship between cash dividends at both the group and existing member levels and the likelihood of sponsoring an IPO indicate that this hypothesis is unlikely to be the sole explanation. Rather, group IPOs appear to be supported by active within group internal capital management of other group members. The same conclusion is supported by an analysis of the other alternative mode of raising equity capital: when an existing group member issues more of its own shares in a seasoned equity offering (SEO), rather than conducting an IPO for its subsidiary. In Column 2, it is clear that the results with respect to the IPO and the SEO decision are not the same. A group is more likely to conduct an SEO when its operating performance is weaker. To be even clearer, we directly compare the IPO decision with the SEO decision by reestimating the conditional logit regression reported in Column 3, excluding all the years when a group does not raise any external equity. The dependent variable is the choice of conducting an IPO or an SEO. If our earlier results are driven by groups opportunistically raising capital 19

22 to take advantage of positive earnings shocks, we should observe that additions to retained earnings have no influence the choice between a SEO and an IPO, conditional on the firm needing to raise capital. However, the results reported in Column 4 show that groups are more likely to conduct an IPO over an SEO when they experience strong operating performance. The control variables in all of our models are in line with the alternative rationales for group expansion. In particular, the coefficients of GRPQ indicate that a group is more likely to conduct an IPO when it has strong growth opportunities. Comparing across groups, the effect of the scale of available group resources, measured by GRPNOFIRMS and GRPFIRMSIZE, both indicate that larger groups are more likely to sponsor IPOs. Also, the more industry concentrated a group is, the greater is the likelihood of an IPO, which indicates that risk sharing is an important motive for group expansion. The significant coefficients for GRPAGE and GRPPRETOPUW suggest that more established groups and groups with established relationships with prestigious underwriters are more active in the IPO market. In terms of control structures, groups that employ dual-class shares are less likely to conduct an IPO (perhaps because they can quite freely raise new capital without concerns about a control loss), whereas the negative coefficient of GRPPYRAMID indicates that there appears to be a limit to the desirability of building a large pyramid with many layers. [INSERT TABLE II HERE] We pursue several other unreported robustness checks. First, we address the possibility that the group s accumulation of retained earnings prior to an IPO within the group is simply due to strong pre-ipo profitability of the newly listed affiliate, which is able go public on its own merits. In this case the group s accumulated retained earnings does not capture the group s ability to support an IPO, but rather is indicative of a strong division of a group 20

23 member going public. To ensure that this explanation is not solely responsible for our results, we collect information on pre-ipo earnings for our group IPO firms from SDC and Worldscope. We remove the earnings attributable to the IPO firm in the year prior to listing and reconstruct the GRPNETINCOME variable. 7 When we re-estimate the first three models reported in Table II with this adjusted measure, we obtain qualitatively similar results. In another unreported test, we check if our finding of a positive relationship between additions to retained earnings and group IPO likelihood is best explained by the internal capital support hypothesis by comparing across the types of group IPOs: those sponsored by an existing listed affiliate (parent) in a pyramid structure and those directly owned by a family group. We re-estimate the regression in reported column 1 of Table II, using only the subsample of group-years associated with a group IPO event of any type. This analysis again documents evidence consistent with the role that a pyramid structure plays in channeling internal capital support: the measures of internal capital accumulation significantly predict the likelihood of observing a pyramid IPO over that of a directly owned IPO. B. Group Member Characteristics That Influence the Likelihood of Becoming an IPO Parent To observe more clearly how groups operate their internal capital markets to support their IPO activity, we next explore which firms within a group are selected to be the IPO s parent firm. The analysis in Table III focuses on firm differences across group members using similar variables to those analyzed in Table II, except that they are calculated at the individual group firm level. In the Column 1 of Table III, we estimate a conditional logit model with group fixed effects and again find that larger additions to a member firm s retained earnings in 7 We are able to construct this variable for approximately 80% of the group IPO firms in our sample. For those with missing pre-ipo earnings information, we exclude their sponsoring groups from the analysis. 21

24 a given year increases the firm s likelihood of becoming the IPO s parent firm. This likelihood also decreases with the size of dividend payouts. In Column 2 of Table III, we substitute group fixed effects with group-year fixed effects to examine which firm within a group in a given year is selected as an IPO parent. Under this stricter test, which only considers cross-member differences in the year associated with a group IPO event (only groups with at least two existing listed firms are considered), we again find that additions to retained earnings significantly raise the likelihood that a particular listed group member is selected to conduct an IPO of one of its subsidiary. In Column 3, we impose firm fixed effects and again find that the likelihood of a given group firm conducting an IPO of a subsidiary rises in financial years with high additions to retained earnings. In Columns 4, we examine for each group firm what explains of the choice of conducting a subsidiary s IPO or an SEO of its own shares, similar to the analysis reported in Column 3 of Table II. The estimates from this model confirm the robustness of previous results with respect to additions to retained earnings. In Column 5, which imposes a further restriction on the group-year sample (requiring each group that is included in the sample to have at least one firm conducting an IPO and another conducting an SEO in a given year), we again use group-year fixed effects and find that the member(s) selected to conduct an IPO tends to have greater accumulation of internal capital than other members that raise funds through an SEO. [INSERT TABLE III HERE] We next investigate whether groups facilitate access to the equity capital market for their new members by supporting them with external equity capital raised through a prior SEO by other existing listed members. It is possible that a group IPO event arises simply because 22

25 the option of undertaking further share issuances by existing members becomes unattractive, perhaps due to the risk of losing control. In an unreported analysis, we re-estimate the regression models reported in the first three columns of Table III after adding an indicator variable for whether an existing group firm has conducted an SEO in the prior three years. We find that this variable does not significantly predict the likelihood of an existing group firm becoming the parent of a new IPO firm. C. Group Members That Hold a Minority Stake in a Group IPO Firm In addition to supporting its own subsidiary to go public, a group firm can assist in the IPO process of subsidiaries of other group members by providing them with intragroup investments. A typical feature of family groups is that member firms do not have just one controlling entity, but also include minority shareholdings by other affiliates. From our IPO ownership data at the time of listing, we identify cases of family group IPO firms in which the group holds multiple stakes (at least 5% of issued shares) and whether in addition to the controlling shareholder, the additional (minority) shareholder can be traced back to another existing listed member in our business group sample. This data allows us to examine the types of group firms that are selected to provide additional funding to a subsidiary about to go public beyond the support of its direct parent. Although the data is limited by the fact that we cannot ascertain the timing of these investments, it still provides some evidence on the flow of internal capital from one group firm to another. In Columns 6 and 7 of Table III, we repeat the analysis reported in Columns 1 and 2 on the alternative dependent variable of the likelihood that an existing group firm holds a minority stake in a new IPO firm of the same group, incorporating group fixed effects and 23

26 then group-year fixed effects. The results indicate that the type of group firms tasked to provide additional investments are those with high cash earnings. Finally, we also examine the factors that explain which groups support their IPOs with additional non-parent equity stakes. In table XX in model xx we estimate a probit model where the depenant is equal to one if non-parent group members hold additional stakes in the IPO and zero otherwise. The results show that group retained earnings play a significant role in explaining which groups are able to provide this additional support. D. Non-family Groups In the last two columns of Table II and Table III, we repeat the main analysis conducted previously for family groups on a separate sample of non-family groups. While the control variables related to the size of a non-family group (a group firm) and the Tobin s Q of a non-family group firm show statistical significance in explaining the decision of these groups to sponsor IPOs, the key internal capital variables are statistically insignificant, both at the group and the parent levels. This suggest that the kind of financial support and coordination provided by family groups for their IPO firms are less present in non-family groups, pointing to the importance of family control at a group s apex that can direct its allocation of internal capital to realize the benefits from using a pyramidal structure. III. Analysis of IPO Market Conditions In this section we examine whether variations in the relative prevalence of business group IPOs across countries and time are consistent with the hypothesis of group financing advantages. This hypothesis predicts that a group s ability to supply internal capital to support an affiliate s external capital raising efforts is more important in countries with high external 24