Equity carve-outs and optimists -Master thesis- Teis Westerhof 988697 November 2014 Supervised by: dr. F. Castiglionesi
Abstract In this paper, I examined the effects of noise traders on the share price of parent companies at announcement date and issue date. I find that at announcement date parents earn an abnormal return of +2.70%, while on issue date they earn an abnormal return of respectively -3.02%. Furthermore, I cannot reject the hypothesis that announcement date abnormal return minus issue date abnormal return is significantly different from zero. I find that positive abnormal returns at announcement date are due to noise traders optimistic about the prospects of the subsidiary who drive up the price of the parent, prior to the IPO of the subsidiary optimistic noise traders had to invest in the parent if they wanted to invest in the subsidiary, therefore the parent s share price also reflects some of this optimism. However, when the subsidiary goes public these optimists will invest in the subsidiary as soon as possible, thereby selling the parent whose share price will drop to pre-optimistic levels, resulting in negative abnormal returns. 2
Table of Contents 1. Introduction... 4 2. Literature review... 11 2.1 The asymmetric information hypothesis... 11 2.2 The divesture gains hypothesis... 15 Refocusing strategy hypothesis... 15 The financing and investment hypothesis... 18 The subsequent event hypothesis... 21 Miscellaneous hypotheses... 22 2.3 Noise trader hypothesis... 23 3. Data... 27 News articles... 30 Calculating abnormal returns... 33 Regression... 34 Hypothesis Development... 39 4. Empirical Results... 42 Abnormal returns... 42 News reports... 44 Regression analysis... 53 Conclusion... 63 References... 66 Appendix... 70 3
1. Introduction An equity carve-out (henceforth: ECO) is a partial public offering of the parent company s stake in its subsidiary for an IPO or rights offering, whereby the parent usually retains the controlling stake in the subsidiary. Although, the parent owns the controlling stake, the two firms are separated and therefore the subsidiary has its own management, board of directors, and will publish its own financial statements. There are several studies examining the effects on stock returns during ECO announcements. All studies find positive abnormal returns (henceforth: AR) for the parent company around an ECO announcement, although the interpretation for this phenomenon differs among the studies. Schipper and Smith (1986) document that an ECO is the only equity financing arrangement undertaken by public firms which shows an increase in shareholder wealth. There are three interpretations regarding the positive abnormal announcement returns, which are the following: The asymmetric information hypothesis, the divesture gains hypothesis, and the noise trader hypothesis. The vast majority of papers about ECOs are about divesture gains, then there is a minority that supports the information asymmetry hypothesis. However, little research has focused on noise trader effects around ECOs. The noise trader hypothesis argues in contrast to the other two hypothesis that ECOs do not create wealth for the parent companies. In this thesis I expand on the noise trader hypothesis, I want to show that the positive ARs for certain divesture types are due to optimistic noise traders rather than wealth creation for the parent. In the current literature some divesture motives earn a significant larger AR than the remainder of the sample. For instance, Allen and McConnell (1998) find that firms that use the proceeds of an ECO to pay down debt earn a significant higher AR than those that retain the proceeds for internal use. In addition Allen and McConnell (1998) find that firms that undertake an ECO are highly-levered and have lower profit ratios, therefore the positive AR could be due 4
to an increased financial position. Or Vijh (2002) who finds that ECOs that mention a reduction in organizational complexity earn a significant larger AR than ECOs that do not mention this rationale for the ECO. However, in contrast to Allen and McConnell (1998) and Vijh (2002) I argue that the formerly mentioned larger ARs are not due to divesture gains but rather noise traders who optimistically misinterpret the information at announcement date, Hand (2006) finds that noise traders optimistically misinterpret information around the ECO, because they mistake an ECO for a spin-off or they misinterpret the accounting gain, usually the market value of the subsidiary is larger than the book value creating an accounting gain in the income statement. I argue that the characteristics of firms that undertake an ECO causes optimistic noise traders to react more positively to these types of divesture rationale. Baker and Wurgler (2006) argue that the propensity to speculate is higher for firms that are harder to value, for instance firms that are young, highly-volatile, in financial distress or have extreme growth opportunities. Therefore, a noise traders can easily defend a wide range of valuations that are in line with their sentiment. I expect that optimistic noise trader have an even greater price impact when the parent is hard to value, such as in the examples of Allen and McConnell (1998) and Vijh (2002) where in first case, the firm is likely to be in financial distress and the latter because the organizational structure is complex making it hard to determine where value stems from. To test the hypothesis that ECOs do not create wealth for the parents and that the larger ARs for certain divesture types are due to the hard to value characteristics of the parent and misinterpretation of information by noise traders, I conduct the research as follows: Firstly, I want to capture the AR around ECO announcements and then subdivided the sample according to the motivation for the ECO, to see whether I can find that some divesture types earn larger ARs than others. Secondly, I want to look at issue date ARs. Hand (2006) finds that parents earn a significant negative AR when the subsidiary goes public. To argue that 5
ECOs do not create wealth for their parents I hypothesize that the announcement date AR minus the issue date AR is not significantly different from zero. To test this statement, I use a two-sample mean comparison test. Furthermore, to determine whether noise traders are optimistic and have a greater impact on hard to value firms I conduct a regression analysis, for the total sample and subdivided according to the news reports, at announcement date and issue date with dependent variable, the AR at announcement/issue date. The independent variables include variables that capture general optimism such as relative size and trading volume and variables that capture the easiness of valuation such as age, volatility, and earnings. Furthermore, I include some variables for rational investors such as firm size, institutional ownership, and dilution. I retrieve data to identify equity carve outs from the Mergers and Acquisition magazine which can be accessed via LexisNexis, and the Dow Jones Factiva. The magazines are available from 1990 onwards. The Mergers and Acquisition magazine yearly published a list of ECOs for the years 1990 until 1999. I supplement the ECOs identified by the Merger and Acquisition magazine by ECOs mentioned in The Wall street journal from 1990 until 2005. To identify event dates I use the Dow Jones factiva, this database has access to news sources and SEC-filing sources, as event date I will use the earlier of the announcement or filing date. The issue date of the carved out company is the first date for which data is available on CRSP. Furthermore, to subdivide the ECOs into different types of divesture rationale, I will use news articles available on the Dow Jones Factiva.To calculate the ARs, I use the CRSP daily database for the company returns and I use the value-weighted market index, the union return of all NYSE, AMEX, and NASDAQ firms, as benchmark return for the market model. In addition, the data regarding the regression analysis is retrieved from CRSP for the security related variables, Compustat for the financial statement items, sec- 6
filings for the expected proceeds and dilution, and lastly the institutional ownership is retrieved from the Thomson Reuters 13f holdings database. I find that the total sample earns a positive AR of +2.70% on announcement date. Furthermore, when I subdivide the sample according to the divesture motivation specified in the news reports I find that they all earn a positive significant return at announcement date. I find that reports that mention financing & investment, complexity, hot industry, regulatory, and management all earn a higher AR than no such reports, however the difference is only significant for reports that mention a hot industry for the subsidiary. The results at announcement date are in contrast to for instance the examples previously named by Allen and McConnell (1998) and Vijh (2002). When looking at issue date return, I find that as opposed to the announcement date, ARs at issue date are negative for the total sample and the subdivided sample. Furthermore their seems to be a relationship between the positive AR at announcement date and the negative AR at issue date, the more positive the announcement date AR the more negative the issue date AR, as I can never argue that announcement date return minus the issue date return is significantly different from zero. These results contradict the conjecture of the divesture gains and information asymmetry hypothesis of wealth creation for the parent when undertaking an ECO, since no AR is earned. The regression analysis shows that the proxies for optimism, trading volume and relative size, are significantly positive and are also able to explain cross-sectional variation at issue date. When looking at the proxies for a hard to value firm, I find that only idiosyncratic volatility has the expected effect on the AR, the larger the uncertainty of stock returns the greater the AR will be, however vice versa at issue date. I find that the coefficients for trading volume relative size and idiosyncratic volatility at announcement dates are equal to minus one times the value at announcement date. Which is caused by noise traders optimistic about the prospects of the subsidiary, who sell the shares of the parent when the subsidiary goes public. 7
Concluding, I do not find that the divesture types earn a larger AR than no such reports, that I cannot reject the hypothesis that announcement date AR minus issue date AR are significantly different from zero, for any sample, that the hard to value proxies have no explanatory power in the AR, and that selling pressure on the parent share cause negative AR. It seems that optimistic noise traders about the subsidiary, drive up the price of the parent and as soon as the subsidiary go public sell parent in favour of the subsidiary. The current literature regarding ARs around ECO announcements can be divided in three views, the first view is the asymmetric information hypothesis which is led by Nanda (1991). He extends the Myers and Majluf (1984) framework, in his framework a consolidated firm has a subsidiary who has a positive net present value investment opportunity. To finance the opportunity the firm can either sell shares of the consolidated firm (parent company shares) or raise cash by selling shares in the subsidiary, by means of an ECO, or the firm can refrain from the project. Nanda (1991) finds that under some circumstances a firm might undertake an ECO, for example when the subsidiary is overvalued and the consolidated firm is undervalued. In this case an ECO reveals information about both the parent s assets and the subsidiary s assets. The positive information effect should dominate the negative information effect of the subsidiary, parent company are expected to have a larger market value than the subsidiary, leading the positive ARs around an ECO announcement. In my regression analysis I find the exact opposite, I find that there is positive linear relationship between the relative size of the subsidiary to the parent and the AR. Moreover, the asymmetric information hypothesis does not explain why parent companies earn on average a negative AR at issue date. In addition to the asymmetric information hypothesis, there is the divesture gains hypothesis. These studies argue that the positive ARs at announcement dates are seen due to various types of divesture gains such as improved efficiency or improved financial conditions. 8
Vijh (2002), finds that ARs for ECOs in different industries earn a significant larger AR than ECOs that operate in the same business segment as their parents. In line with Desai and Jain (1999) who show that firms that increase their focus by means of a spin-off have superior long-run ARs compared to a corresponding sample of non-focus increasing spin-offs largely explained by an improved operating performance. Allen and McConnell (1998) argue that firms undertaking an ECO are likely to be capital constrained, they find that prior to an ECO parents have significantly higher debt ratios, lower interest coverage ratios, and lower profitability. Therefore, a positive AR is expected due to an improved financial position following the ECO. However, in contrast to Vijh (2002) and Allen and McConnell (1998) I do not find that these firms earn a significant larger AR at announcement date. Moreover, I find negative ARs at issue date and when subtracting the issue date AR from the announcement date AR, I cannot reject the hypothesis that they are significantly different from zero. In conclusion, when taking issue date in consideration parents do not earn an AR and thus parents do not benefit from divesture gains. A completely different view is the noise trader view, who argue that the positive ARs around ECO announcements are due to optimistic noise traders. Noise traders optimistically misinterpret the new information (Hand, 2006). For instance, noise traders may misinterpret the accounting gain in net income. Hand (2006), provides three arguments for his noise trading statement. Firstly, he finds that parent s earn a significant positive AR of +2.3% at announcement date, however when the subsidiary goes public, parent s earn a significant negative AR of -2.4%. Secondly, he finds that proxies for noise trading based on information known at announcement date are also able to explain 11% of cross-sectional variation in AR at issue date, under market efficiency the proxies should not be able to explain variation at issue date. Thirdly, he finds that the regression coefficients, for noise trading at issue date are individually and jointly equal to minus one time those observed at announcement date. 9
Therefore, Hand (2006) concludes that noise traders optimistically react to an ECO announcement and that this reaction is corrected by rational investors at the issue date. In contrast to Hand (2006) and in accordance with Lamont and Thaler (2003), I find that, optimistic noise traders are substituting parent shares for subsidiary shares as soon as the subsidiary goes public, therefore negative issue date ARs are due to selling pressure on the parent rather than a correction by rational investors. In conclusion, I do not find support for the asymmetric hypothesis I find that the larger the relative size of the subsidiary to the parent the larger the AR at announcement date will be. Furthermore, I do not find support for the divesture gains hypothesis. Firstly, when I divide the sample according to different divesture motives for the ECO stated in the news reports I do not find that a specific motive earns a larger AR than the remaining cases. Secondly, when incorporating the issue date in the analysis I am not able to reject the hypothesis that announcement date ARs minus issue date ARs are significantly different from zero, hence there seems to be no AR to be earned by parents regardless of the motive stated. I find that Lamont s and Thaler s (2003) argument hold noise traders optimistic about the prospects of the subsidiary drive up the price of the parent, prior to the IPO of the subsidiary, optimistic noise traders had to invest in the parent if they wanted to invest in the subsidiary, therefore the parent s share price also reflects some of this optimism. However, when the subsidiary goes public these optimists will invest in the subsidiary as soon as possible, thereby selling the parent whose share price will drop to pre-optimistic levels. The remainder of this thesis is as follows. Section 2 is the literature review. Section 3 is the data section and describes the sampling method, methodology and descriptive statistics, section 4 contains the empirical results and lastly section 5 is the conclusion. 10
2. Literature review 2.1 The asymmetric information hypothesis The asymmetric information hypothesis is led by Nanda (1991). However, also Schipper and Smith (1986) mention information asymmetry and both base their analysis on the Myers and Majluf (1984) model. In the Myers and Majluf (1984) model, a firm has one asset in place and one investment opportunity. To finance this opportunity the firm must issue common stock otherwise the investment opportunity is forgone, in addition the firm cannot make use of debt. The other assumptions are the following: Firstly, the managers of the firm have more knowledge about the assets in place and the investment opportunities, and that both managers and investors realize there is a case of information asymmetry. Secondly, they assume that financial markets are perfect and efficient, thus no taxes, transaction costs, or other market imperfections plus all publicly available information is incorporated in the share price. Thirdly, it is assumed that the investors are passive and management acts in the interest of the old shareholders. In this model, Myers and Majluf (1984), show that management might not want to issue new shares to invest in a positive net present value project, because it harms the old shareholders, and hence the firm suffers from underinvestment. Myers and Majluf (1984) suggest that one way to mitigate this information asymmetry effect is to separate the financing of the assets in place and the investment opportunity. Schipper and Smith (1986), argue that if the subsidiary has an investment opportunity one way to get separate financing is to undertake an ECO. They, argue that as opposed to a seasoned equity offering (which experiences negative ARs around announcement dates, e.g., Asquith and Mullins (1986), find that seasoned equity offerings on average experience a 11
negative AR of -2.7% 1 in a two day period after the announcement is made.) an ECO has a claim on the cash flows of the subsidiary only and this way the information asymmetry between the managers and potential investors might be alleviated. Information asymmetry might be alleviated if management has less private information about the subsidiary than about consolidated firm, if this is the case ECO should experience less negative ARs around the announcement. Furthermore, Schipper and Smith (1986) predict that under two circumstance an ECO is expected to experience positive market returns. The first case, if information about the investment opportunity is made publicly without negative implications for the non-subsidiary part of the firm. The second case, if separate financing leads to management not refraining future positive net present value projects. In this context an ECO can be seen as a way to finance growth opportunities. Nanda, (1991) wants to construct a common framework that could reconcile the findings of Schipper and Smith (1986), positive market returns around an ECO announcement and negative market reactions in case of other forms of equity financing. Nanda (1991), does this by extending the Myers and Majluf (1984) framework. In the model that Nanda (1991) constructs there is a publicly traded parent company that owns a wholly owned subsidiary, the subsidiary has a positive net present value investment opportunity. Once the firm has received the project it decides whether or not to raise external capital to finance the project. The firm can either sell stock of the consolidated firm, raise cash by undertaking an ECO, or do nothing and refrain from the investment opportunity. In addition the same assumptions apply as in the Myer and Majluf (1984) paper. Nanda (1991) finds that some firms want to initiate an ECO, when the subsidiary is overvalued and the consolidated firm is undervalued. Therefore, Nanda (1991) states that: by their financing decisions, firms reveal information not just information about the value of assets in place of the subsidiary but also about the value of the assets in 1 Masulis and Korwar (1986) examine a sample of 690 seasoned equity offerings and find a negative abnormal return -3.25%. 12
place in the rest of the corporation (p.1719). Based on his findings Nanda (1991) argues that in this case, rational investors will raise the stock price as positive information hits the market, the consolidated firm being undervalued, and the positive effect on the larger parent dominates the negative effect on the subsidiary, subsidiary being overvalued, leading to positive ARs for the parent. Furthermore, Nanda (1991) argues that firms that want to raise cash by selling stock of the consolidated corporation are likely to be overvalued, possibly explaining the negative ARs around seasoned equity offerings. Although, Nanda (1991) constructs a model that possibly might explain positive ARs around ECOs announcements, he does not provide observational evidence for his statement. Krishnaswami and Subramaniam (1999) look at information asymmetry and spinoffs, they find that companies before the spinoff have higher levels of information asymmetry than their industry peers and information asymmetry is largely mitigated by the spinoff. Furthermore, Krishnaswami and Subramaniam (1999) find that ARs are positively related to information asymmetry, i.e., the higher the information asymmetry, the higher the gains around a spinoff. Since, an ECO is a partial spinoff, Krishnaswami s and Subramaniam s (1999) results might be analogous to ECOs. Powers (2003), provides some observational support for the information asymmetry hypothesis by ECOs. He finds that operating performance of the carved out firms is better than the operating performance of the peer group at the time of the carve out. However, Powers (2003) finds that over a period of five years operating performance becomes equal to the industry mean. Moreover, Powers (2003) finds that percentage of equity sold to the public is negatively related to the performance of the subsidiary, i.e., the more the parent sells, the worse the future performance. He argues that if management knows that the current operating performance of a subsidiary is not sustainable but the market does not, the parent is likely to carve out the subsidiary and sell a large part of its equity, i.e., the subsidiary will sell above its intrinsic value. Thus, Powers (2003) 13
concludes that an ECO is a seemly way to sell overvalued assets and generate cash. A study by Slovin, Sushka, and Ferraro (1995) provides some direct evidence for the asymmetric information hypothesis. They find that rivals of subsidiaries experience a significant -1.11% drop in stock price, presumably because the industry is overvalued. When looking at rivals to parents Slovin et al. (1995) find they earn the normal return. However, Hand (2006) finds that rivals to both the parent company and the subsidiary on a stand-alone basis are not significantly different from zero. Therefore, he argues that the market does not consider the industry both to the parent and the subsidiary to be undervalued or overvalued. In addition to Hand (2006), Hulbert, Miles, and Woolridge (2002) find that rivals of the parent company experience negative market returns, also contradicting the information asymmetry hypothesis. Furthermore, Hulbert et al. (2002) argue that their results support the divesture gains hypothesis. Since, according to the divesture gains hypothesis rivals should experience negative market returns as the parent company becomes more competitive, for instance Allen and McConnell (1998) find that parent companies lower their relative to the industry high leverage. The difference in results might be explained by the sample used, Slovin et al. (1995), only look at a sample of 36 ECOs whereas Hand (2006) and Hulbert et al. (2002) look at a more representative sample of respectively 265 and 185 ECOs. In addition to Hand (2006) and Hulbert et al. (2002), Vijh (2002) also rejects Nanda s (1991) information asymmetry hypothesis. According to Nanda s (1991) model, as mentioned before, a firm undertakes an ECO when the parent is undervalued and the subsidiary is overvalued. Since, Nanda (1991) assumes the parent is larger than the subsidiary the positive effect of the parent dominates the negative effect on the subsidiary. Following this logic, if the subsidiary is getting larger the positive ARs should become smaller and start to get negative when the subsidiary makes up more than one half of the consolidated firm. Vijh (2002) conducts an empirical test of subsidiary size relative to non-subsidiary assets of the parent 14
firm. He shows that in cases where the subsidiary assets are smaller than the non-subsidiary assets of the parent the average AR is +1.19%, and in cases the subsidiary assets are larger than the non-subsidiary assets the average AR is +3.73%. These results are the exact opposite of what the information asymmetry model predicts. Furthermore, he divides the subsidiaries into 10 subsets of relative size and finds that ARs are increasing with relative size, again the exact opposite of Nanda s (1991) theory. In addition, Allen and McConnell (1998) find that large ECOs, those whose book value of subsidiary divided by book value of the firm before the ECO lie above median earn an AR almost twice as large as those whose size is below median, +7,68 against +3,78 for the smaller ECOs. The empirical observations do however not fully resemble the assumptions made in asymmetric information. Nanda (1991) assumes a perfect capital market and an all-equity firm while in reality firms there are market imperfections such as taxes and firms have varying amounts of debt, also it is hard to determine the market value of the subsidiary before the announcement date. Vijh (2002) tries to overcome these problems by using net assets, and assumes the market value of liabilities is equal to the book value, and the market value of the subsidiary on the listing date. In conclusion, there is little evidence supporting the information asymmetry hypothesis of Nanda (1991) 2.2 The divesture gains hypothesis A prominent paper about divesture gains is the paper written by Vijh (2002), he combines and examines, different divesture rationale for ECOs. He constructs hypothesis on the basis of other studies, the first hypothesis is the refocusing strategy hypothesis. Refocusing strategy hypothesis Comment and Jarrell (1995) examine the effects of increased focus on shareholder wealth. They find that adding or subtracting a business segment results in a 5% shareholder wealth change. An explanation Comment and Jarrell (1995) provide for this negative economies of 15
scope is that those economies of scope for conglomerates are unused. Furthermore, Comment and Jarrell (1995) find that the best average long-term share performance is seen in firms that increase their focus and the worst average performance is seen in firm that increase their performance with a net difference of 7.4% over a 34 month period. John and Ofek (1995) show that firms that engage in asset sales experience a significant improvement in operating performance in the three years following the asset sale, and besides the improved efficiency companies also have positive ARs around sales announcement. Furthermore, John and Ofek (1995) find that in 3 out 4 cases the asset sold is in an unrelated business with the seller s assets, hence there is an increase in focus. John and Ofek (1995) argue that the improvement in operating performance after the asset sale is due to better management of the assets in place. Daley, Mehrotra, and Sivakumar (1997), examine spin-offs, spin-offs are different from asset sales in the sense that asset sales usually involve an exchange of cash whereas a spin-off do not. A spin-off is the creation of an independent company whose shares are distributed to the company that spins-off the subsidiary. Furthermore, a spin-off is different from an asset-sale as, amongst others, Schleifer and Vishny (1992), argue that firms often engage in asset-sales when they are financially distressed. Hence, positive returns around announcement of asset-sales might come from financial restructuring, since spin-offs do not involve cash announcement returns are free of this bias. Daley et al. (1999), find that focus increasing spin-offs, the spun-off company is in another industry than the former parent company, earn a positive AR of +4.3% while spinoffs in the same industry as the former parent earn a positive AR of +1.4%, although insignificant. Thus, supporting the hypothesis that corporate focus maximizes shareholder wealth. Desai and Jain (1999) look at the long-run stock returns for firms that undertook a spin-off and find that focus increasing spin-offs earn positive ARs over a three year period whereas non-focus increasing spin-offs earn negative ARs (although insignificant). Moreover, 16
Desai and Jain (1999) find that the change in focus is positively related to a change in operating performance, they argue that a focus increasing spin-off reduces the diversity of assets and thereby increasing the efficiency of management. The number of ECOs undertaken coincide with the trend toward increased focus in the 1980s and 1990s (Vijh, 2002). In addition Vijh (2002), finds that firm that undertake an ECO operate in more business segments and industries than the average firm. Vijh (2002) does not find that firms mentioning refocusing earn a significant higher AR than firms that do not mention a typical strategy for the ECO. However, Vijh (2002) does find that subsidiaries that operate in a different industry than their parent earn a higher abnormal than subsidiaries operating in the same industry, difference in AR is +1.54%. Otsubu (2013) looks at the Japanese market and his results are similar to those of Vijh (2002) although the difference is smaller. Otsubu (2013) finds that ECOs undertaken in different industries earn a higher positive abnormal than ECOs in the same industry as the parent, +1.76% against +1.55%. In addition, Vijh (2002), looks at what happens after an ECO and he finds that ECOs that announces they turn into a spin-off earn a higher AR than ECOs without a second event announcement, the difference is large +2.61%. However, it must be noted that an equity carve-out cannot always be seen as a form of divesture, because in most cases the parent company retains quite a large stake in the carved out subsidiary, for instance Hand (2006) finds that in his 265 ECOs sample the median stake of the parent held in the subsidiary is 75.7%. Furthermore, carved out subsidiaries are often reacquired by their parents, for instance Otsubu (2009) shows that in his sample around 14% of the carved out companies are reacquired, or Vijh (2002) who finds that around 20% of carved out companies are reacquired. 17
The financing and investment hypothesis An important paper that looks at proceeds of an ECO to be used for financial or investment purposes is written by Allen and McConnell (1998). While most other studies view an ECO as an issue of equity, Allen and McConnell (1998) view an ECO as the sale of an asset, to raise funds to finance activities such as paying off debt or to finance investment opportunities. They base their work on the Lang, Poulsen, and Stulz (1995) study. Asset sales are associated with positive ARs (e.g., Hite, Owers, and Rogers (1987) find that asset sales are associated with a positive AR of +1.66% for the seller and +0.83% for the buyer.), the explanation Hite et al. (1987) provide is improved efficiency of assets due to a better allocation. Lang et al. (1995) provide another explanation, they argue that managers value firm size and control even if it hurts shareholders, such that manager do not want to sell assets for the sole purpose of improved efficiency. They, argue that for these types of management a manager is likely to sell asset only when the firm needs financing and external financing is too expensive to pursue its objectives. Lang et al. (1995) name this view the financing hypothesis of asset sales, and they argue that the sale of an asset in this case can be seen as good news since management is only willing to sell an asset if the price is right. Their empirical results support their hypothesis, they find that firms that sell assets suffer from poor performance and/or have high leverage. Furthermore, Lang et al. (1995) find that the year before the sale of assets, the firms on average do not have a net income significantly different from zero, supporting their theory that firms engage in an asset sales only when it is driven by financial needs. Furthermore, Lang et al. (1995) find that the use of proceeds is especially important to shareholders, if the firm announces that it will use the proceeds to pay off debt, stock prices are significantly positive but if on the other hand the firm announces that it will retain the cash within the firm the reactions are negative, although this result is insignificant. Lang et al. (1995) argue that announcement returns for firms withholding cash are zero due to agency 18
costs and managerial discretion that comes with a larger sum of cash, although their results do not support this theory. Thus, the findings of Lang et al. (1995) are inconsistent with the hypothesis of improved efficiency stated by amongst other Hite et al. (1987), otherwise one would see positive returns for the cash withholding firms as well, and in addition also the more profitable firms would sell a part of their assets since not only firms with low profitability are likely to hold comparatively inefficient assets. Allen and McConnell (1998) apply the financing hypothesis of Lang et al. (1995) on ECOs, although they admit that an ECO has certain characteristics that distinguish a carveout from a plain asset sales. Different characteristics are; an ECO is a sale of assets to the public rather than to one buyer, ECOs are undertaken for the sole purpose of raising funds, and other than with a sale of assets the parent company (selling party) keeps an interest in the subsidiary (asset) and thus still has some control over the asset (Typically, the parent holds 80%. Allen and McConnell (1998)). However, an ECO is similar to a sale of assets in a way that the proceeds can be kept in the firm or distributed to debt holders or shareholders. Allen and McConnell (1998) make the presumption that managers only undertake an ECO when management has to fund activities and has to engage in an ECO because it is otherwise capital constraint, otherwise it would for example issue debt because management values control, which is partly lost in an ECO. This presumption predicts that prior to an ECO, a firm suffers from poor operating performance and therefore is paled to issue stock to finance new investment opportunities as there would be little demand for new shares, or the firm is high levered and used its debt capacity. Allen and McConnell (1998) find that in their sample firms prior to an ECO had significantly lower interest coverage ratios, higher debt ratios, lower profitability ratios than their industry peers. These results are consistent with the presumption of a capital constraint firm. Furthermore, Allen and McConnell (1998) divide their sample into two subsamples according to the use of the proceeds. The first sample is the payout 19
sample, if a firm falls in this sample, the firm distributes the proceeds of an ECO to the financiers. The second sample is the retention sample, firms in this sample use the proceeds for internal purposes such as capital expenditures, investments in working capital, or acquisitions. They find that all firms in the sample earn a positive AR on announcement of +2.12%. However, when they divide this in two subsamples of distribution to financiers or cash retained in the firm, Allen and McConnell (1998) find that on average a positive AR of +6.63% is earned if cash is distributed and a negative AR of -0.01% if the cash is retained in the firm. The negative AR supporting the managerial discretion and agency costs hypothesis of lang et al. (1995). Otsubu (2013), examines Japanese ECOs, which are different from ECOs in the United States in the fact that parents usually retain the capital relationship with the carved out company as opposed to ECOs in the United States. Furthermore, ECOs are more common in Japan, Japanese companies tend to have many publicly listed subsidiaries, this is different for US firms, for example Allen and McConnell (1998) argue that US firms only undertakes an ECO when it is financially constrained. Otsubu (2013), argues that since the parent subsidiary relation does not change that much, wealth is created, Otsubu finds that the Japanese stock market reacts positively, AR of +1.67%, on an ECO announcement, by something else than the relationship between the parent and the subsidiary. Therefore, Otsubu (2013) looks at the financial aspect of an ECO. Otsubu (2013), shows that parent companies that undertake an ECO suffer from financial constraints and tend to decrease their leverage after an ECO is undertaken. Furthermore, he finds that market reacts positively to an ECO when the parent company is highly levered. These results, suggest that the market values the financial restructuring of the parent company and thus supports the results of Allen and McConnell (1998). 20
The subsequent event hypothesis Klein, Rosenfeld, and Beranek (1991) document that ECOs are often followed by a second event, they find that an ECO is often followed by a complete spinoff or by a reacquisition by the parent firm. Furthermore, Klein et al. (1991) find that ECOs that turn into a complete selloff are associated with positive ARs and those that are reacquired are associated with insignificant returns. Perotti and Rossetto (2007), argue that an ECO is a way to gain information from the market about how it values the subsidiary, upon this information the parent firm can either decide to spin-off of the company or the reacquire the subsidiary, since Perotti and Rossetto (2007) argue that an ECO is a temporary event. Perroti s and Rossetto s (2007) theory might explain why there more ECOs in industry with larger uncertainty, high sales growth, and large R&D investments. (Allen & McConnell, 1998) Otsubu (2009) examines the relationship between the ECO and the secondary event. Otsubu (2009) finds that when the parent-subsidiary relationship is retained the ECO announcement have a positive market reaction as opposed to a separate relationship which is not associated with positive market returns. Therefore, Otsubu (2009) argues that the ARs around ECO announcement dates are due to a combination of the carveout itself and the subsequent event. The subsequent event being either M&A activity, reacquisition, spin-off, or secondary offering. Where M&A activity refers to the parent selling its stake in the subsidiary to a third party, secondary offering means a sale of parent stake in the subsidiary to the public, spin-off means distribution of subsidiary shares to the parent firm s shareholders, and reacquisition is the reacquisition of all subsidiary shares by the parent. Otsubu (2009) finds that the market reacts positively to all four secondary events, although the results for M&A activity are insignificant. Otsubu (2009) argues that this is due to the market expecting M&A activities as secondary event at the ECO announcement since it has the greatest benefit. According to Otsubu (2009) it exploits divesture gains, it exploits gains of a package sale: 21
Share blocks are often sold at a premium to exchange prices (Otusubu, 2009), and it exploits gains of exploitation, the parent firm is more likely to sell shares of the subsidiary if they are higher than they expected (Otsubu, 2009). The other types of secondary events do not have all of these three types of gains, but still have positive announcement returns. Gleason, Madura, and Pennathur (2006) find that ECOs that are reacquired experience a negative or insignificant market return at the time of the ECO announcement. Furthermore, Gleason et al. (2006) document that carveouts that are reacquired perform substantially worse than carveouts that are not reacquired and that both the subsidiary and the parent earn positive ARs when the parent announces the reacquisition, ARs are even larger when the two businesses are in the same industry. Gleason et al. (2006) argue that the positive ARs around reacquisition announcements might be seen because the parent firm knows more about the potential of the subsidiary and since its performance was worse than for firms with other second events, the subsidiary might not have been able to live up to its potential. Their findings suggest that the subsidiary is being undervalued by the market, as opposed to Nanda (1991) his theory of ECOs on average being overvalued. In addition to their results contradict the findings of Klein et al. (1991), probably because Gleason et al. (2006) use longer period and therefore have a larger sample base. Vijh (2002) also looks at the secondary event and similarly to Gleason et al. (2006) finds positive ARs for carved out subsidiaries that are reacquired. Furthermore, Vijh (2002) finds that companies that completely spin-off earn the highest AR and companies with no secondary event earn the lowest AR, which is in line with the refocusing hypothesis. Miscellaneous hypotheses Schipper and Smith (1986) argue that the possibility to invest in the carved out company alone, increases the demand for information of the subsidiary and possibly leading to a better understanding of the subsidiary, which might lead to a higher value of the subsidiary by the 22
market. Krishnaswami and Subramaniam (1999) find that complexity is reduced by undertaking a spin-off, they find that earnings forecast of the two separate firms are more accurate than of the consolidated firm. Furthermore, Schipper and Smith (1986) find that in 14 out of 76 ECOs a better understanding of the subsidiary is provided as important argument to undertake the ECO. In addition Vijh (2002) finds that news reports that mention a reduction in complexity earn a significant 2.34% higher AR than all other cases. The results contradict the asymmetric information theory which argues that firms undertake an ECO when the subsidiary is overvalued not when it is undervalued. Another explanation for wealth enhancement given by Schipper and Smith (1986) is improved managerial incentives, they argue that many ECOs are associated with restructuring of managers responsibilities and incentives. For example the remuneration of the manager is now tied to the share price of the subsidiary, giving a bigger incentive to managers ultimately leading to a higher value. Aron (1991), examines managerial incentives around spin-offs and she finds that even the possibility of a spin-off in the future improves the subsidiary s management current incentives. Management will already act as if it is being monitored although performance is only observed when the spin-off actually takes place. Lastly, Vijh (2002) argues that there might be a few other reasons to undertake an ECO, those being: regulatory compliance, facilitating mergers, and tax reduction. 2.3 Noise trader hypothesis The asymmetric hypothesis and the divesture gain hypothesis both assume the market is efficient and therefore the ARs around announcement dates can be seen as wealth creation for the parent s shareholders. However, Hand (2006) argues that the positive ARs around an ECO announcement are due to noise traders, who optimistically misinterpret the new information (Hand, 2006). For instance, he suggests that ECOs are often mistaken as spinoffs, who have proven to be shareholder wealth increasing (Vijh, 1994). Also, noise traders may mistakenly 23
interpret the accounting gain in an ECO as the book value is usually lower than the market value. To support his statement Hand (2006) provides three arguments. Firstly, He finds that on the issue date of the subsidiary parents experience a negative AR of -2.4% in contrast to a positive announcement AR of +2.3%. Usually the parent retains a majority stake in the subsidiary (median stake is around 72%, Vijh (2002)), therefore subsidiary returns and parent returns are clearly related, thus the negative AR on announcement date could be due to the subsidiary being overpriced (Hand, 2006). However, Hand (2006) finds that 68% of the subsidiaries are underpriced suggesting that the negative AR at issue date does not stem from the link between the parent s and subsidiary s return at issue date. Schmeling (2007), finds that individual sentiment and subsequent market movements are negatively correlated, which could explain the negative AR at issue date, overoptimistic noise traders drive up the price, away from intrinsic value, which has to be corrected at some point, possibly the issue date. Lamont and Thaler (2003) argue that this phenomenon could possibly be explained by investors who desire to hold the subsidiary drive up the parent s price and then when the subsidiary goes public sell the parent s shares and buy the subsidiary s shares. However, the efficient market hypothesis assumes that demand and supply shocks do not affect share prices, as there is no change in fundamental value (Scholes, 1972). Therefore, if this explanations holds the market is inefficient at announcement date and/or issue date. According to the efficient market hypothesis changes in fundamental value occur due to new information that hits the market, the positive ARs at announcement date suggest that positive news hits the market, and issue date ARs vice versa. However, since it is known at announcement date that the subsidiary goes public these negative ARs violate the efficient market hypothesis principle, since public information known at announcement date cannot be expected to earn an AR beyond the announcement date (Hand, 2006). These three explanations suggest that the market is inefficient at either announcement date or issue date, 24
or the market is inefficient at both dates. If the market is inefficient prices may deviate from fundamental value and thus the ARs seen at announcement date may not create wealth for the parent s shareholders as is suggested in the asymmetric information hypothesis and divesture gains hypothesis. Secondly, Hand (2006), constructs proxies for noise traders including ECO size, abnormal trading volume, and listing on the different exchanges. He finds that at announcement dates his proxies are able to explain 18% in cross-sectional variation in the parents ARs at announcement date, and 11% of cross-sectional variation in ARs at issue date. The cross-sectional variation on issue date is all based on information which was public on the announcement date, according to market efficiency all information should be priced in at announcement date and therefore this proxy should not explain eleven percent but zero percent. Thirdly, hand (2006) finds that the regression coefficients, for noise trading at issue date are individually and jointly equal to minus one time those observed at announcement date., suggesting that mispricing is corrected at issue date. Therefore, given these three pieces of support Hand (2006) concludes that noise traders optimistically react to an ECO announcement and that this reaction is corrected by rational investors at the issue date. Another paper that looks into noise trader demand and ECOs is written by Lamont and Thaler (2003), they examine 18 ECOs that are followed by complete spin-offs in the tech industry. They find that in six cases the efficient market assumptions are clearly violated, that is in those cases with a negative stub (i.e., where the market value of the parent is less than the market value of its ownership in the subsidiary.). A famous example of a negative stub value is the 3Com and Palm case. In this case 3Com carves 5% of the shares of Palm out and the remaining 95% is intended to be distributed to 3Com s shareholders as a stock dividend. 3Com shareholders receive 1.5 shares of Palm for every share of 3com they own. Therefore, the price of 3Com must be at least 1.5 times as much as the price of palm since the price of 25
3Com shares can never be negative. On the first listing date, the price of Palm jumped to $95.06 and the price of 3Com fell to $81.81, hence 3Com had a negative stub value of -$60.7 ($81.81-1.5*95.06). Here the law of one price is clearly violated and therefore the market is not efficient, at least at that time. Rational arbitrageurs could theoretically make a profit by buying a share of 3Com and sell short 1.5 shares of Palm pocket $60.7 and at the time of distribution of the Palm shares use them to cover the short position and sell the share in 3Com, total profit will be $60.7 plus the share prices of 3Com. However, there are some frictions that limit arbitrage, the biggest friction is imperfect information and the cost to get an understanding of the exploitable arbitrage opportunity, especially when there is little to examine such, only 82 negative stub cases between 1985 and 2000 (Mitchell, Pulvino, Stafford, 2002). In addition, in the case of Palm there is only a 5% free float, therefore there is limited supply for borrowing making it costly to short the stock. Hand (2006) finds in his study that mispricing is not as extreme as in the Palm case however he argues that it is maybe the same underlying optimism that creates this mispricing. Bayar, Chemmanur, and Liu (2011), develop a theory of investor optimism and ECOs, in addition they provide an explanation of the existence of negative stub values in ECOs. In the model of Bayar et al. (2011) a firm has a project and the firm has to finance to project by either selling stock in the consolidated firm or undertaking an ECO. Furthermore, they assume investors have heterogeneous beliefs and are short sale constraint. Bayer et al. (2011) argue that under some circumstances the firm chooses to undertake an ECO. Firstly, if there is heterogeneity among investors, this way the firm can lower the cost of external financing by selling equity to the most optimistic investors. Secondly, if the project is perceived more optimistic by investors than management, also lowering the cost of external financing. Thirdly, if the project is appealing to a different investor base than the parent firm, a different investor base will likely have different beliefs about the project, in this case the investor base 26