Department of Finance, School of Business, The George Washington University, 2201 G. Street, NW, Washington, DC 20052, USA c

Transcription

1 On the Acquisition of Equity Carve-Outs Chintal A. Desai a,*, Mark S. Klock b, Sattar A. Mansi c a Department of Economics and Finance, College of Business Administration, The University of Texas-Pan American, 1201 West University Drive, Edinburg, TX , USA b Department of Finance, School of Business, The George Washington University, 2201 G. Street, NW, Washington, DC 20052, USA c Department of Finance, Insurance, and Law, Pamplin College of Business, Virginia Tech, Blacksburg, VA 24061, USA Abstract This version: May 24, 2011 Published in 35 Journal of Banking and Finance (2011), pp We examine the role played by the parent s motive in undertaking a carve-out; the parent s post-ipo influence over the carved-out subsidiary; and anti-takeover provisions and industry structure of a carve-out on its acquisition likelihood and its acquisition premium. We find that the probability and hazard of a carve-out acquisition increase when the parent s objective is to unlock the value of a subsidiary and when the parent and the subsidiary are tied with a product-market relationship. We also find that the post-ipo parent ownership significantly affects the acquisition likelihood and the level of acquisition premium. Additional analyses examining the post-ipo carve-out status suggest that the product-market relationship and post- IPO parent ownership increase the probability of re-acquisition. JEL classification: G34; G32 Keywords: Equity carve-outs; Acquisition; Parent-subsidiary; Corporate restructuring * Corresponding author. Tel.: ; fax: addresses: desaica@utpa.edu (C. Desai), msk20@gwu.edu (M. Klock), smansi@vt.edu (S. Mansi). Acknowledgement The authors would like to thank the anonymous referee, the editor (Ike Mathur), Gregory Elliehausen, Bharat Jain, Gergana Jostova, Irvin Morgan, Christo Pirinsky, Robert Savickas, Emanuel Zur, and seminar participants at Florida State University (College of Law), George Washington University, Ohio University, University of Texas-Pan American, Midwest Finance Association (2007), and Southern Finance Association (2009) for their comments and suggestions. Ditina Desai contributed by proofreading the manuscript. This paper is partially based on Desai s Ph.D. dissertation at The George Washington University. A portion of this research was completed while Desai was at University of Maryland, College Park. All remaining errors are the authors responsibility. 1

2 1. Introduction An equity carve-out (ECO) is a form of corporate restructuring whereby the parent firm offers a subset of its subsidiary s common shares to the public. 1 Research evidence indicates that ECOs are temporary arrangements for the parent firms and the majority of ECOs are acquired within a few years after their initial public offerings (IPOs). 2 In such a scenario, do we know a priori the factors which can predict the eventual acquisition of an ECO? And do these factors have an impact on the ECO shareholders wealth, when it actually gets acquired? To our knowledge, these questions have not been explored in the corporate restructuring literature and we attempt to fill this gap. Specifically, we analyze the effects of: the parent s motive for conducting a carve-out; parental control on the post-ipo carve-out; anti-takeover provisions in the charter of a carve-out; and industry structure of the carve-out on the likelihood of an acquisition and the level of acquisition premium. Within the likelihood of acquisition, we look into the probability of acquisition and the hazard of acquisition. We also assess the impact of our chosen variables on the subsamples of ECOs that are either re-acquired by the parent firm or acquired by a third-party. The prior work of other scholars facilitates our research endeavors. For example, Perotti and Rossetto s (2007) theoretical model implies that the likelihood of an ECO acquisition increases when the parent s motive for conducting a carve-out is to unlock the value of its subsidiary. As per the authors, prior to the carve-out, the parent is unaware of the fair value of its synergy with the subsidiary, which precludes it from making an informed decision to keep either full or no ownership in the subsidiary. Subsequent to the carve-out, market participants 1 For a review on ECOs and other forms of corporate restructuring, see Eckbo and Thorburn (2008). 2 For example, Klein et al. (1991) show that 44 of 52 ECOs in their sample are either re-acquired or sold off to a third-party by the parent firms. 2

3 assist the parent firm in ascertaining the value of its synergy with the subsidiary. If the synergy is positive, then the parent firm decides to re-acquire the carved-out subsidiary, else if the synergy is negative, then the parent decides to sell it off to a third-party. Based on the authors insights, we posit that, if the parent s motive is to unlock the value of its subsidiary, it is likely to be interested in determining its synergy with the subsidiary. Therefore, this carved-out subsidiary s acquisition is imminent, either by the parent itself or by a third-party. An equity carve-out differs substantially from other forms of corporate restructuring such as spinoffs and asset sales in terms of the post-divestiture parent role in the subsidiary. 3 Allen and McConnell (1998) show that, on average, the parent retains 69% of the ownership in the subsidiary. The ownership in an ECO provides required voting rights to the parent, which allows it to appoint its nominee on the board and on the audit committee of the carve-out. 4 The parent firm, generally, appoints its past employees to key executive positions of the carve-out. In addition, prior to a carve-out transaction, a parent can sign long-term business agreements with the subsidiary, which allows it not only to preserve the existing synergy but also to exert undue influence over the subsidiary. 5 We hypothesize, as explained in Section 2.2, that the post-ipo parental control can have an effect on the ECO s acquisition. 3 In spinoffs, the parent firm distributes its ownership in the subsidiary on a pro-rata basis to its existing shareholders, whereas in the case of an asset sale, the parent firm sells its ownership in the subsidiary to a third-party on a private negotiation basis. In both cases the parent retains no ownership in the post-divesture subsidiary, suggesting minimal parent control. However, a few papers show the significant role of the parent in the spun-off subsidiary. For example, Wruck and Wruck (2002) show the involvement of the parent in appointing the topmanagement team of the spinoff. 4 Throughout the paper, the term ownership refers to the voting ownership rather than cash flow ownership since the former can have a significant impact on the acquisition event. The difference between the two is relevant only for equity carve-outs with dual class shares. 5 The long-term agreements with the parent firm can prevent an ECO from obtaining a better deal in an arm s-length transaction. For example, the Appendix of Atanasov et al. (2010) provides the case of Coca-Cola s (KO) influence over its partially-owned (49% ownership) subsidiary Coca-Cola Enterprises (CCE) through long-term business agreements. 3

4 The anti-takeover provisions and the industry in which an ECO operates can also play a role in its acquisition. Field and Karpoff (2002) find that the probability of acquisition of an IPO decreases when it employs takeover defenses in its charter. Takeover defenses might also affect the level of acquisition premium; though the evidence is mixed. For example, Megginson (1990) finds an increase in premium for firms with anti-takeover provisions, while Field and Karpoff (2002) do not find any significant relation between the level of premium and anti-takeover provisions. Regarding the relevance of the industry, Perotti and Rossetto (2007) conjecture that ECOs operating in high-technology (hi-tech) industries have a longer time-to-acquisition due to a higher level of valuation uncertainty in those industries. In contrast, hi-tech firms also offer changes in technology and above-market returns which can increase the probability of their acquisition (Kohers and Kohers, 2000). Using a sample of 222 equity carve-outs (of which 137 are acquired) for the period of 1986 to 2004, we find that the likelihood of acquisition increases when the parent s motive for undertaking an ECO is to unlock its value, when the parent retains majority ownership in the post-ipo carved-out subsidiary, and when the parent and the subsidiary are tied with productmarket relations. We also find some evidence that when the ECO operates in the hi-tech industry, the likelihood of acquisition increases. Further, we find an inverse relation between the level of acquisition premium and whether the parent ownership in the ECO is above 50% prior to the acquisition. In terms of the differences between the parent re-acquisition and a third-party acquisition, we find that the probability of re-acquisition increases when the parent s motive is to unlock value, when the parent retains majority ownership in the post-ipo carve-out, and when the parent and ECO have an ongoing product-market connection. 4

5 Our results are robust to self-selection and endogeneity issues. Michaely and Shaw (1995) find that the bigger and less profitable parent firms are likely to divest their subsidiaries by conducting carve-outs instead of spinoffs. This choice of going public in the first place is likely to induce self-selection bias in our estimation. To address this, we perform a two-stage Heckman analysis for the probit models. In the first stage, we analyze the choice between a carve-out and a spinoff, and in the second stage, given that the parent has selected to perform an ECO, we analyze its likelihood of acquisition. In terms of endogeneity, the parent might be anticipating the ECO s future acquisition at the time of consummating the carve-out transaction, and accordingly decides the post-ipo ownership in the carve-out. In that case, the retained ownership in the carve-out is likely to be endogenously determined. We address this issue by performing an instrumental variable probit analysis and bivariate probit analysis, which are similar to that of a two-stage least squares analysis. In the first stage, we explain the parent s post-ipo ownership in a carve-out, and in the second stage, we use the predicted value of the parent retained ownership for explaining the probability of an acquisition. We do not report these results in this paper to economize on space, but they are available in Desai et al. (2011). Our paper contributes to the corporate restructuring literature in at least four ways. First, we undertake a comprehensive study by collectively analyzing the IPO and acquisition events of a carve-out. 6 Second, we assess the likelihood of acquisition of carve-outs by analyzing the probability and the hazard of acquisition. The latter part, to our knowledge, has not been investigated. Third, we explore the impact of contributory factors, other than the parent retained 6 Several studies seek to establish a link between the IPO and the acquisition events of a carve-out. For example, Klein et al. (1991) show that the probability of re-acquisition increases with the increase in post-ipo parent ownership. Hulburt (2003) finds that the parent firms of the ECOs that were subsequently acquired experience greater wealth gains at the time of carve-out announcements. Otsubo (2009) shows that the market participants anticipate a third-party acquisition and accordingly adjust the share price of the parent on a carve-out announcement. 5

6 ownership, such as the parent motives for conducting a carve-out, parent-subsidiary business relations, anti-takeover provisions, and the industry in which the carve-out operates on the carveout s acquisition. Lastly, we explain differences in the likelihood of acquisition and the acquisition premium for the subsamples of parent re-acquisitions and third-party acquisitions. The remainder of the paper is organized as follows: Section 2 illustrates our hypotheses, and Section 3 highlights the data, variable measurements, and descriptive statistics. Section 4 provides the empirical results. Section 5 concludes the paper. 2. Hypotheses development 2.1. Reasons for conducting a carve-out and its acquisition When conducting a carve-out, the parent firms incur a substantial amount of explicit costs such as legal, underwriting, and filing fees. They also underprice in order to successfully consummate the offering; Hogan and Olson (2004) show a first-day return of 13%. In addition, there are costs due to the disincentive effect (Perotti and Rossetto, 2007). Before a carve-out, the operating profits of the wholly-owned subsidiary belong entirely to the parent firm. After a carve-out, the parent firm incurs costs in operating the partially-owned subsidiary, but the profits are pro-rata distributed to the shareholders which also include minority shareholders. Then, what are the benefits for the parent firm in conducting an equity carve-out? Schipper and Smith (1986) describe three main reasons for conducting a carve-out: financing, restructuring, and unlocking the value. The financing reason involves the instances when the parent conducts an ECO to finance its own or its subsidiary s growth projects, to discharge parent debt, or to raise funds to meet capital expenditure requirements. The restructuring reason involves the instances when the parent carves out the subsidiary to gain from 6

7 the divestiture. 7 Finally, the unlocking the value reason involves the cases when the parent undertakes an ECO for investors to better understand the operations of the carve-out. At the time of an IPO and subsequent to it, the subsidiary needs to comply with the Securities and Exchange Commission (SEC) by filing forms such as registration statements, proxy statements, 10-Ks, and annual reports. From these filings, the investors can obtain the relevant information about the carved-out subsidiary. As mentioned earlier, the theoretical insights of Perotti and Rossetto (2007) help establish a possible linkage between the parent motive at the time of a carve-out and the subsequent acquisition of the carve-out. They argue that the status of a subsidiary as a fully-integrated and an independent identity are two extremes, and it depends on the synergy between the parent and the subsidiary. By selling out full ownership in a subsidiary, the parent firm carries an irreversible loss if it finds a positive synergy with the subsidiary in the future. At the same time, by keeping the subsidiary fully-integrated, the parent firm is unable to assess the value of the synergy. An ECO of the subsidiary provides the parent a better alternative. On the one hand, carving out the subsidiary at the first stage causes the parent firm to incur costs related to the public issue and the loss of operational efficiency. On the other hand, it helps the parent firm determine the value of its synergy with the subsidiary more accurately while retaining control (through a majority ownership) in the subsidiary. Therefore, the carve-out of a subsidiary provides the parent firm with an indefinite American style option to buy back the remaining ownership from the minority shareholders or to sell out its ownership to a third-party. 7 The divestiture gains hypothesis posits that the carve-outs help the parent firms improve investment efficiency, focus on their core businesses, and reduce agency costs by aligning managers and shareholders interests through writing effective contracts (Hulburt et al., 2002; Vijh, 2002). 7

8 The exercise of this option depends on the value of the underlying synergy between the parent and subsidiary. These synergies (operational, financial, etc.) can be positive or negative. The positive value of the synergy indicates that the parent is better off by keeping 100% ownership in the carved-out subsidiary. Thus, the parent can achieve synergistic gains such as lower transaction costs than what it would gain in arm s length transactions, and it can avoid losses due to the disincentive effect. The negative value of the synergy implies that the parent is better off by removing the carved-out subsidiary from its portfolio. Therefore, the parent may decide to sell its ownership to a third-party. We conjecture that if the parent s motive is to unlock the value of a subsidiary, then its intentions are likely to ascertain its synergy with the subsidiary, and it would like to possess the real option of re-acquiring or selling off the remaining portion of the subsidiary. Therefore, our first hypothesis is: H1 : The parent s motive of unlocking the value of the subsidiary suggests a higher likelihood of the ECO s acquisition Post-IPO parental control and the acquisition of a carve-out In this subsection, while forming the hypothesis on the relationship between the post-ipo parental control and the acquisition of the carved-out subsidiary, we mainly refer to two mechanisms by which the parent can exert its undue influence over the carve-out. These are parent retained ownership and its ongoing business connections with the carve-out. The parent firm obtains a real option to re-acquire or to sell off the remaining portion by retaining the majority ownership in the carved-out subsidiary (Perotti and Rossetto, 2007). The majority ownership also helps the parent firm obtain the approval of the ECO board with less effort should it decide to re-acquire the subsidiary, because through majority ownership the 8

9 parent can appoint its nominees to the carve-out s board. Even in the case of a third-party acquisition, the bidder effectively needs to convince only one shareholder (the parent firm) to tender shares, which increases the likelihood of acquisition. In addition, majority ownership creates less-dispersed ownership in the carve-out, which helps mitigate the free-rider problem as pointed out in Grossman and Hart (1980). Majority ownership also allows the parent firm to appoint its past employees to the key management positions in the carve-out. These top executives can potentially affect the outcome of a major corporate event such as acquisition. Therefore, the majority ownership in a post-ipo carve-out suggests an increase in the likelihood of a carve-out s acquisition. Further, the property rights theory of Grossman and Hart (1986) and Hart and Moore (1990) suggests that when Company A and Company B have a business relationship and Company A has incentives to control assets and resources of Company B, then Company A will maintain ownership in Company B. For ECOs, based on the predictions of the property rights theory, Boone (2003) finds that the parent firms maintain high ownership in carve-outs if they have a product-market or financial relationship. Therefore, we conjecture that if the parent and subsidiary are linked through long-term business relationships such as product-market and financial, then the parent firms are likely to retain majority ownership in the subsidiary which will increase its likelihood of acquisition. Therefore, our second hypothesis is: H2 : Majority ownership by the parent firm in the post-ipo carve-out and ongoing business connections between the parent and the ECO increase the likelihood of acquisition of the ECO. The parent through its control over the subsidiary can impact the level of acquisition premium. In the case of re-acquisitions, it is reasonable to claim that the parent may expropriate 9

10 wealth of the minority shareholders. However, the incentive to expropriate minority shareholder wealth will be weaker when the parent retains a larger fraction of the ECO because a smaller fraction of any premium paid will accrue to the minority shareholders. 8 Additionally, the parent firm s re-acquisition bid needs to be approved by a special committee of independent directors who do not have any vested interest with the parent firm. Therefore, the difference in the acquisition premium between a third-party acquisition and a parent re-acquisition might be insignificant. Indeed, prior empirical findings suggest mixed views. On the one hand, Slovin and Sushka (1998) analyze a sample of parent-subsidiary mergers that also includes mergers of equity carve-outs. As per their study, for the period , the subsidiary announcement day return is 19% for 105 parent re-acquisitions; whereas that statistic is 21% for 38 third-party acquisitions, and the difference between the two is statistically insignificant. On the other hand, Atanasov et al. (2010) show the instances when the parent firms deliberately transfer risk to the publicly-traded subsidiaries and expropriate wealth from the minority shareholders. These mixed findings preclude a decisive hypothesis about the minority wealth expropriation in the case of equity carve-outs, and we rely on our data and analyses to guide us on this Anti-takeover provisions in a carve-out and its acquisition Managers adopt anti-takeover provisions in the charter of a firm mainly to avoid the external market for corporate control. 9 Ambrose and Megginson (1992) show that the probability of acquisition decreases when the firm uses blank-check preferred stock a form of an antitakeover provision. Megginson (1990) reports that the trading price and the acquisition premium 8 We thank the referee for bringing this important point to our attention. 9 For a review on different forms of anti-takeover provisions, see Gompers et al. (2003). For the impact of antitakeover provisions on the performance of a spin-off parent firm, see Chemmanur et al. (2010). 10

11 of the superior voting shares are higher than those of the otherwise equivalent class of restricted voting shares. In the case of IPOs, Field and Karpoff (2002) show that the probability of acquisition of an IPO decreases when it has anti-takeover provisions in its charter at the time of the IPO. Therefore, our next hypothesis is: H3 : Anti-takeover provisions in the charter of an ECO decrease the likelihood of its acquisition, and increase the level of acquisition premium. 3. Data and variables 3.1. Data description We obtain the initial sample of 619 equity carve-outs, offered over the sample period of 1986 to 2004, from the Security Data Corporation (SDC) s new issues database and issues of the Mergers and Acquisitions magazine. 10 From this sample, we eliminate carve-outs for which parents/carve-outs are mutual trusts, limited partnerships, foreign firms, finance firms, limited liability firms, and joint venture firms. We also exclude observations which are unit offerings, target stocks, spinoffs, and for which data are missing from the Center of Research in Security Prices (CRSP) tapes for the parent/carve-out. This screening reduces our sample to 295 carveouts. Finally, we eliminate carve-outs for which the parent firms have already announced plans to divest their remaining ownership 11 (44 carve-outs) and the data are missing in the (Q-data) microfiche (29 carve-outs). This yields a final sample of 222 carve-outs from 199 parent firms. It contains wholly-owned (163 carve-outs) and partially-owned (59 carve-outs) subsidiaries. 10 The cutoff year of 2004 allows us to follow all carve-outs for at least a three-year period, which provides sufficient time for parents to fully divest or re-acquire their subsidiaries should they choose. 11 In these cases, the parent firms generally announced plans to divest their remaining portion within a year of IPO by tax-free spinoffs. For more on such carve-out/spinoff combination transactions, see Thompson and Apilado (2006). 11

12 For each carve-out, we collect the parent firm s reason for conducting a carve-out from news articles in LexisNexis. The ownership, board composition, top management team, and business relationship data are from SEC filings including proxy statements, 10-Ks, and prospectuses. We collect merger announcement dates and merger related details such as offer price, exchange ratio, and winning bidder from news articles in LexisNexis and Factiva, and available proxy statements before the acquisition. In order to avoid errors due to discrepancies in the Standard Industrial Classification (SIC) codes in the Compustat and CRSP databases (Kahle and Walkling, 1996), we collect SIC codes primarily from LexisNexis and SEC filings, and the missing ones from SDC, Compustat, and CRSP, in that order. We obtain offer size data from Factiva, SEC filings, and SDC. Finally, we extract data on the time-to-acquisition, size of the carve-out, and size of the parent firm from CRSP Dependent variables For the logit analysis on the probability of acquisition, we follow Field and Karpoff (2002) and assign a value of one to a carve-out that was acquired within five years from its IPO date, and zero to a survived one. For the hazard analysis, we define Acquisition Time as the number of months between the last and the first trading dates of a carve-out as given in CRSP. For carve-outs which are actively traded as of December 31, 2007, we take this date as the last available trading date. For the analysis on acquisition premium, we use three approaches to compute the level of premium. First, we consider the possibility of an information leakage prior to the merger announcement (Schwert, 1996). The variable Acquisition Premium_42 is the ratio of the price offered per share of a carve-out to the carve-out s price per share on the 42 nd trading day prior to 12

13 the merger announcement, minus one. For the stock-to-stock exchange, we compute the carveout offer price using the available exchange ratio and the bidder s price on the announcement day. For five acquired carve-outs, since we do not find the price on the 42 nd trading day prior to the announcement, we use the price on the sixth trading day prior to the announcement. For the second and the third approach of computing the acquisition premium, we follow Moeller (2005). We define Acquisition Premium_Returns as the buy and hold (excess) return during the merger announcement date. Specifically, we compute the carve-out s buy and hold return for 11 trading days starting with the five trading days prior to the merger announcement and subtract the buy and hold return on the CRSP value-weighted index for the same time period. Finally, we define Acquisition Premium_06 which is similar to Acquisition Premium_42, except that we compute the ECO price on the sixth instead of 42 nd trading day prior to the merger announcement Explanatory variables We group our explanatory variables into five sets of variables. The first set pertains to the motive of the parent firm to undertake an equity carve-out. The second set consists of factors through which the parent can exert control over the post-ipo carve-out. The third category contains our proxies for the anti-takeover provisions in a carve-out and its industry structure. The fourth category represents the set of control variables. Finally, the fifth category contains the set of variables which are related to a merger transaction. The variable definitions, their detailed description, and the data sources are available in the Appendix Factors related to the parent motive 13

14 We extensively review the news reports at the time of the IPO announcements and identify three categories of reasons for conducting an equity carve-out. These are financing/investing, restructuring, and unlocking the value. 12 These are not mutually exclusive categories due to the cases when the parent provides multiple reasons for conducting a carve-out. There are also other instances when the reason for conducting a carve-out does not fall into any of these three categories, for example, the reason of conducting a carve-out due to apartheid concerns or to meet government regulations on an antitrust lawsuit; we note these cases under the other category. First, if a parent undertakes a carve-out for general corporate purposes, for working capital, for repayment of debt, for capital expenditures, for possible acquisitions, or for taking out the parent from financial distress, we consider these instances in the financing/investing category. Second, we define the restructuring reason as the instances when the parent consummates the carve-out transaction to focus on core business or as part of a larger restructuring plan. Finally, the category of unlocking the value consists of the cases when the parent undertakes a carve-out to create a liquid market for the subsidiary shares, to allow the market to value the subsidiary, or to create a liquid market for employee shareholders Factors related to the post-ipo parental control over the carve-out We study the parental role in the post-ipo carve-out via equity ownership, board and audit committee composition, top-management team, and long-term business relationships. The 12 For more detailed examples of the language used by parent firms when they announce carve-outs for unlocking the value, see Desai et al. (2011); Chemmanur and Liu (2011); and Nanda and Narayanan (1999). 13 We also define the parent s motive to unlock the value by considering only those announcements in which the parent undertakes an ECO to allow the market to value the subsidiary. Our results remain qualitatively similar with this definition. 14

15 most direct way the parent firm can exert control over the ECO is by retaining majority ownership. We also define the reduction in parent ownership as the difference of parent ownership before and after the IPO; a smaller value of this variable indicates a higher level of parental control over the post-ipo carved-out subsidiary. To capture the non-linear relation between the parent ownership and our dependent variables, we consider two indicator variables reflecting whether the parent retains above 50% and above 80% ownership. The ownership allows the parent firm to appoint its personnel as directors and topexecutives of the carve-out firm. Directors in the carve-out can be one of three types: insider, outsider, or parent representative. Parent representative directors are current or past employees of the parent firm. Following Boone (2003), we define the top-management team of a carve-out being affiliated with the parent firm if the chairman, CEO, or president of a carve-out is either a current or past employee of the parent firm. We also measure the parent influence over the audit committee of the carve-out by identifying whether any of the parent s past or current employees serve on that committee. We define a product-market relation between the parent firm and the ECO when they have buyer-supplier, technology sharing and research and development, marketing and distribution, or specialized service agreements (Boone, 2003). Similarly, we define a financial relationship when the parent and the ECO are involved in a long-term financial agreement through letters of credit, bank guarantees, or term loans (Boone, 2003). We also consider industry relatedness of the parent and subsidiary to capture the possibility of parental involvement in the subsidiary if they both are in the same industry Factors related to anti-takeover provisions and industry structure of a carve-out 15

16 We use three types of anti-takeover provisions. These are whether a carve-out has a staggered board, two classes of shares, and whether its state of incorporation is Delaware. We restrict our analysis to only three anti-takeover defenses due to unavailability of (a) IPO prospectuses for the ECOs prior to the year 1996 and (b) takeover defenses on standard databases such as the Investor Responsibility Research Center (IRRC) for an ECO firm because of its small size. Regarding the industry structure of a carve-out, we assign a dummy variable that takes on a value of one if the firm operates in the hi-tech sector (SIC code of 28, 357, 36, 38, or 737) Other control variables Governance related: We measure ownership of the CEO, insiders (executives and directors), and blockholders. Because ownership data are not reported when the insider/ceo owns less than 1% and the blockholder owns less than 5% of the outstanding shares, we use both continuous and indicator variables for insider/ceo and blockholder ownership. Regarding the directors other than the parent directors, insider directors are the employees of the carve-out but have no relationship with the parent firm. Outside directors are not the employees of the carveout or of the parent firm. We measure board size as the number of directors immediately after the IPO. Board size is considered large when the number of directors is above the median board size. We capture the CEO power by measuring whether she is also a chairperson of the board. Size and performance related: We control for the relative size (ratio of the carve-out size to the parent firm size), IPO offer size, and carve-out stock market performance in our analysis. 14 We follow the definition of a hi-tech industry in Benou et al. (2008), who in turn rely on the SDC. An example of a hi-tech carve-out is Networth Inc (SIC: 3577) which went public on 11/25/1992 and is in the business of computer peripheral equipments. Its parent, Tandem computers (SIC: 7377), also operates in a hi-tech industry. 16

17 Factors related to a merger transaction For analysis on the level of acquisition premium, we also control for variables that are related to a merger transaction. These are whether a lawsuit is filed during the merger process, the ratio of bidder size to target size, the carve-out size prior to its merger, whether more than one bidder attempts to acquire the carve-out, and whether the transaction is financed through only cash. All these factors are likely to impact the acquisition premium Descriptive statistics In Panel A of Table 1, we report the incidence of carve-outs based on offer year and post- IPO parent firm s ownership. We classify carve-outs into three categories based on the CRSP delisting codes as of December 31, 2007: survived, acquired, and failed carve-outs. Survived carve-outs with a delisting code of 100 represent 18% of the sample and are still actively traded in the market. Acquired carve-outs with a delisting code between 200 and 290 represent 62% of the sample and are acquired by either the parent firm (22%) or a third-party (40%), and trading activities ceased after the acquisition date. Failed carve-outs represent the remaining 20% of the sample (46 carve-outs); of which 44 carve-outs are delisted (delisting code between 500 and 591), one is liquidated (delisting code between 400 and 490) and one is exchanged for another class of common stock (delisting code of 331). [Insert Table 1 about here] In terms of the year-wise distribution of the sample, the majority of ECOs are in the years 1996, 1991, and 1986, with 22, 21, and 20 carve-outs, respectively. The subsample of acquired 17

18 ECOs display that 17, 15, and 14 transactions are in the years 1986, 1996, and 1987, respectively. In the period from 2001 to 2004, there are only seven carve-outs in our sample. 15 We also examine the incidence of carve-outs for the parent ownership after the IPO. We divide the sample into three categories: those with post-ipo parent ownership less than 50%, between 50% and 80%, and more than 80%. The largest number of deals occurred in the ownership category between 50% and 80% (about 37%), followed by the greater than 80% category (about 34%), and the less than 50% category (about 29%). The parent firms retained majority (above 50%) ownership in 77% of acquired carve-outs, whereas, in survived carve-outs that statistic is only 49%. This difference of parent ownership in survived and acquired carveouts suggests a possible relation between parent ownership and carve-out acquisition. Similar to the findings of Klein et al. (1991), 92% of the re-acquired carve-outs in our sample have parent ownership above 50%, which suggests that the probability of re-acquisition of carve-out increases with the post-ipo parent ownership. In Panel B of Table 1, we show the incidence of carve-outs and parent firms based on industry using two-digit SIC codes. Although there are 199 parent firms which conducted a total of 222 carve-outs, for simplicity we consider a combination of one parent firm and one carveout. The majority of the ECOs operate in the manufacturing industry (41%) and service industry (27%). Also, these industries significantly represent the acquired carve-outs sample with 41% and 31% of the sample, respectively. Based on our sample, 72% of the carve-outs operating in the service industry and 62% operating in the manufacturing industry are eventually acquired. Among the parent firms, 49% and 18% of sample firms operate in the manufacturing and service sectors, respectively. 15 During this period, we exclude 13 observations of carve-out/spinoff combination in which the parents have already indicated the intent to divest their remaining portion. 18

19 In Panel A of Table 2, we report summary statistics of variables pertaining to post-ipo parental influence on the carve-out and our chosen controls for the total sample. The parent ownership prior to the carve-out has a mean of around 95% and median of 100%, whereas parent ownership subsequent to the carve-out has a mean of 62% and median of 68%. The high post- IPO parent ownership in a carve-out suggests that the parent firm has an interest in controlling the subsidiary, which might not be possible through other forms of restructuring such as spinoffs and asset sales. 16 The parent representative directors constitute, on average, 42% of the carve-out board, which indicates the possibility of the parent firm s role in the operating and strategic decisions of a carve-out. Blockholder ownership has a mean of around 7%. Similarly, insider and CEO ownership in a post-ipo carve-out on average are around 4% and 2%, respectively. The number of directors serving on the carve-out has a median of seven. The carve-out offer size has a mean of $174.5 million, a median of $48.5 million, and minimum and maximum values of $2.5 million and $8.7 billion, respectively. The median size of a carve-out is 22% that of the parent firm. For our sample firms, the average first year excess return over the value-weighted CRSP index return has a mean of -3.5%, which shows that in the short-run the ECOs tend to underperform. [Insert Table 2 about here] Panel B of Table 2, we report summary statistics of dummy variables that proxy for the parent s motive and its influence, anti-takeover provisions, industry structure and other control variables. In around 40% of our sample firms, the reason for conducting a carve-out is for financing related or for investment related activities. We also note that in around 11% of our 16 We also notice that in 13 out of 222 carve-outs the parent retained no ownership after the IPO of the subsidiary. In four of these carve-outs, the parent and subsidiary are tied through product-market agreements, and in none of these 13 carve-outs, the parent s motive for undertaking a carve-out is to unlock the subsidiary s value. 19

20 sample firms the parent conducts a carve-out to unlock the subsidiary s value. In 157 out of 222 equity carve-outs, parent ownership is above 50%; the board-chairperson and CEO are associated with the parent firm in 75% and 51% of carve-outs, respectively. These numbers suggest the parent firm's involvement in the operating and strategic decisions of the publiclytraded subsidiary. Regarding the business relationships, 59% of the carve-outs in our sample are linked to their parent firms through product-market agreements and 75% are linked through financial agreements. These statistics on business relationships suggest an alternative mechanism, other than ownership, by which the parent firm can control the subsidiary (Atanasov et al. 2010). In 33% of our sample firms, the ECO and parent have the same two-digit SIC code. In our sample around 41% (18%) of ECOs use a staggered board (dual class shares). We also observe that 69 carve-outs operate in the hi-tech industry, representing 31% of our sample. Although the parent firms generally retain significant ownership in the ECO, we also observe a notable presence of blockholder, insider (executives plus directors), and CEO ownership in 37%, 54%, and 27% of our sample firms, respectively. Finally, in 91 out of 220 carve-outs, the CEO is also the chairperson of the board. Out of 222 carve-outs, we could not obtain data on audit committee, business relationship, chairperson s affiliation with parent, and CEO duality for 22, one, four, and two carve-outs, respectively. 17 [Insert Table 3 about here] In Table 3, we report statistics similar to those reported in Table 2 for the sample of acquired carve-outs. As shown in Panel A of Table 3, the mean and median time-to-acquisition are 69 and 54 months, respectively. The minimum and maximum time-to-acquisition is eight and 217 months, respectively. Out of 137 acquired carve-outs, 48 are re-acquired by the parent firm, 17 In empirical analyses, we conservatively assign values of zero for these missing data; results remain the same without such adjustment. 20

21 and 89 are acquired by a third-party. The average and median time-to-acquisition in the case of parent re-acquisitions are 59 and 48 months, respectively, whereas those statistics are 74 and 58 months, respectively for third-party acquisitions. These differences suggest that the reacquisitions tend to occur sooner than the third-party acquisitions. The average and median acquisition premium, based on the price of the ECO on the 42 nd trading day prior to the merger announcement date, are 46% and 37%, respectively. However, when we compute the premium using the price of the target on the sixth trading day prior to the announcement those statistics are 34% and 25%, respectively; this lower level of premium suggests a possible information leakage prior to the merger (Schwert, 1996). The median post-ipo parent ownership in acquired carveouts is 72%. As shown in Panel B of Table 3, around 77% of acquired carve-outs have majority parent ownership. We also observe that around 56% of the acquired carve-outs have parent ownership above 50% prior to their acquisition. In our sample of acquired carve-outs, 31% have the parent and carve-out in the same industry and 16% have dual-class shares. Finally, 58% of carve-out mergers in our sample are financed through only cash, and in around 11% of mergers a lawsuit is filed either by a competing bidder or by the minority shareholders. 4. Results 4.1. Likelihood of acquisition: Probability of acquisition In Table 4, we report the results of the logit analysis in which the dependent variable Acquired takes on a value of one if the carve-out is acquired and zero if the carve-out is traded in the market ( survived ), within five years of its IPO offer date. 18 For this analysis, following 18 Because our cut-off date is 12/31/2007, we could not follow one carve-out (IPO offer date 11/20/2003) for the full 21

22 Field and Karpoff (2002), we ignore the carve-outs which are delisted or have gone bankrupt ( failed ) within five years of their IPOs (24 observations). Since we are restricting this sample to carve-outs that are acquired within five years of IPOs, our overall sample is of 198 carve-outs, of which 77 are acquired. Our primary specification (Model 1) is: ey i Acquired i = 1 + e y i (1) Where, y i = A 0 + A 1 Unlocking i + A 2 Parent Own i + A 3 (Hi Tech) i [Insert Table 4 about here] As reported in Model 1, the coefficient of on the dummy variable Unlocking indicates that when the reason for undertaking a carve-out is to unlock the value, the odds of that carve-out to get acquired increase by a factor of 2.97 over the instance when the reason is other than to unlock the value. In terms of the probability, if the reason for undertaking a carve-out is to unlock the value, then the probability of acquisition for that carve-out changes from 0.5 to 0.75; this change is significant at the 5% level (z-value: 2.28). 19,20 The coefficient of on Parent Own suggests that an additional one-percentage decline in the post-ipo parent ownership in a carve-out causes a reduction of 0.791% in the probability of acquisition of this carve-out. 21 The coefficient of on Hi-Tech shows that if the carve-out operates in the hifive years. 19 More precisely, the increase in odds is equal to e = Assuming that if there are reasons other than to unlock the value, then the probability of acquisition p 1 = 0.5, and the odds = 1. When the reason is to unlock the value, the new odds = 2.97, which will make new p 1 = In all our multivariate analyses, t and z values are based on heteroskedasticity-robust standard errors clustered at the parent firm level. 21 More precisely, assuming that before the IPO, the parent owns 100% of the subsidiary. In the first case, the parent retains 81% (i.e. it sells 19% of its ownership in IPO) of the subsidiary, and in the second case, the parent retains 80%. Thus, Parent Own in the first (second) case is 19% (20%), and the change in Parent Own is equal to 0.01, and the new odds = e = Therefore, the new probability of acquisition is , which was previously

23 tech industry, then there is approximately a 66%(34%) chance that this carve-out will be acquired(survived). This result supports findings of Kohers and Kohers (2000) who conjecture that the likelihood of acquisition will be higher in the hi-tech industry due to above-average returns in that industry. In Models 2 to 5, we change our primary specification to incorporate other factors of interest and controlling variables. As mentioned in Section 2.2, the parent is likely to retain majority ownership in the carved-out subsidiary if it has business relations with the subsidiary. In that case, the effect of post-ipo parent ownership may confound the effect of ongoing business connections on the carve-out s acquisition. Therefore, to separate these two effects, we exclude Parent Own and include variables related to the business relationships in Model 2. The coefficient of on dummy variable Product Market indicates that the odds of acquisition increase by a factor of 1.85 when the parent and subsidiary are connected through a productmarket relationship, and this change is significant at the 10% level (z-value: 1.79). In Models 4 and 5, we use a dummy variable Parent Own 50% instead of a continuous variable Parent Own. As shown in Model 4, the coefficient of on Parent Own 50% suggests that the odds of ECO acquisition increase by a factor of 2.38 when the parent retains majority ownership in the post-ipo carve-out, and this difference is significant at the 10% level. In Models 2 and 5, we include variables Financial and Same Industry. For brevity, we do not show their coefficients, due to their statistical insignificance. As shown in Model 4, the coefficient of on Dual Class indicates that the odds of acquisition increase by a factor of 2.50 when the carve-out has dual class shares compared to the case when it has shares with only one class. This change is significant at the 10% level. In addition, in unreported regressions, when we add the variable Dual Class in our Models 1 to 3, 23

24 we find its insignificant impact on the likelihood of acquisition in Models 1 and 3, and in Model 2, its impact is significant at the 10% level. This counterintuitive finding to our hypothesis H3, at first, suggests that the anti-takeover provisions actually increase the likelihood of acquisition in the case of carve-outs. One plausible explanation for this finding is: in examining the SEC filings, we find that in the case of ECOs with dual class shares, the parent firm generally retains full ownership of the superior voting shares while it retains minimal or no ownership of the inferior voting shares. In this manner, the parent voting ownership increases significantly whereas its cash flow ownership is negligible. We define the ownership based on the voting rights. Therefore, this effect of dual class shares is possibly the effect of parent ownership in the post-ipo carve-out. 22 If we examine the effect of another form of anti-takeover provision (Staggered) in Models 3 to 5, it is negative albeit statistically insignificant. Therefore, we do find some support of our hypothesis H3. In an unreported analysis, we also test the difference between the estimated coefficients of various categories of parent motives. For example, in Model 2, the test of difference of estimated coefficients for Unlocking and Financing has a χ 2 value of 6.02 with a p-value of Similarly, the estimated coefficient for Unlocking is significantly higher than that of the Restructuring, with a χ 2 value of 6.74; whereas there is no significant difference in the estimated coefficients of Financing and Restructuring (χ 2 = 1.82). These results suggest that the probability of acquisition significantly increases when the parent s motive is to unlock the value of the subsidiary compared to that when the motive is either to finance or to restructure the operations. 22 For example, in our sample of 39 ECOs with dual class shares, the average parent ownership is 82.6% and for 183 ECOs without dual class shares, that statistic is 57.7%; the difference is significant at the 1% level. 24

25 Overall, the results reported in Table 4 indicate that the probability of ECO acquisition increases when the parent s motive for conducting a carve-out is to unlock the value, with the increase in parent ownership in the post-ipo carve-out, and when the parent and subsidiary are connected through a product-market agreement. These findings support our hypotheses H1 and H Robustness check Our results of the logit analysis on the ECO acquisition can potentially produce biased results if we ignore the fact that at the time of making the subsidiary a public company, the parent could choose whether to conduct a carve-out or to conduct a spinoff. In addition, if the parent decides to conduct a carve-out, then it can determine the level of ownership it needs to retain in the carved-out subsidiary. If we ignore the former, i.e., the choice between a carve-out and a spinoff, our results are likely to be biased due to self-selection. And if we ignore the latter, i.e., the possibility that the parent ownership in the post-ipo carve-out is endogenously determined, our results may be incorrect due to endogeneity. To check our results for sample selection bias we merge our carve-out sample with a sample of spinoffs and perform a two-step Heckman procedure. To check our results for endogeneity bias, we repeat our analysis using exogenous instrumental variables. Our results remain the same under both sets of analyses and there is no indication that our findings are affected by selection or endogeneity bias. Details of these tests are in Desai et al. (2011) Likelihood of acquisition: Time-to-acquisition 25