TARGET STOCK PRICE RUNUP PRIOR TO ACQUISITIONS. Matthew David Brigida. A Dissertation Submitted to the Faculty of. The College of Business

Transcription

1 TARGET STOCK PRICE RUNUP PRIOR TO ACQUISITIONS by Matthew David Brigida A Dissertation Submitted to the Faculty of The College of Business in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy Florida Atlantic University Boca Raton, FL December 2009

2 Copyright by Matthew David Brigida 2009 ii

3

4 ACKNOWLEDGEMENTS I wish to thank my loving family, particularly my wife Kathleen, parents David and Susan, grandparents Cacilda, Olga and Walter, Aunt Cherie, and Uncles William and Barry. iv

5 ABSTRACT Author: Title: Institution: Matthew David Brigida Target Stock Price Runup Prior to Acquisitions Florida Atlantic University Dissertation Advisor: Dr. Jeff Madura Degree: Doctor of Philosophy Year: 2009 Information leakage before full acquisitions has been widely documented. The information leakage, and the resulting pre-bid runup in the target s stock, generally increases the total cost of the acquisition. That is, information leakage and the ensuing pre-bid runup is a gain to the target and loss to the acquirer. Herein, I first ascertain the characteristics of full acquisitions that affect the amount of information leakage. I find that if the acquirer borrows to finance the acquisition then information leakage is greater. Further if the acquirer is foreign, if the target is a high-tech firm, and if the target has options on its stock all increase information leakage. I find hostile deals are effective in reducing information leakage. Lastly, information leakage increases in the percentage of managerial ownership. I next hypothesize that the identity and intent of partial acquirers is known to market participants before the announcement of a partial acquisition. I find that the market can anticipate whether a partial acquirer intends to fully-acquire or take an active role in the management of the target. Also, the market anticipates whether the acquirer is a private investment find or a non-financial corporation. Further, the v

6 acquirer s identity or intent is fully reflected in the target s stock price before the announcement of the partial acquisition. These results help explain why there are few partial acquisitions as precursors to full acquisitions. I next hypothesize that macroeconomic factors affect information leakage, and may serve as a signal of when to speculate on acquisitions. I find that information leakage is positively related to shocks in both expected economic conditions and financing costs, the latter signaling to speculators that acquisitions are imminent. I also find information about an imminent full acquisition is leaked earlier when there are positive shocks to economic conditions and financing costs. vi

7 TARGET STOCK PRICE RUNUP PRIOR TO ACQUISITIONS List of Tables...viii List of Figures. ix Essay 1: Characteristics of Runup Before Full Acquisitions...1 Introduction....1 Contribution...2 Literature Review...3 Empirical Investigation of Factors Affecting the Magnitude of Runup....7 Data and Methodology...34 Results..44 Conclusion...78 Essay 2: Target Runup Before Initial Schedule 13D & 13G Filings Introduction Contribution. 83 Literature Review.88 Empirical Investigation Data and Methodology Results Implications Conclusion.140 Essay 3: The Temporality of Runup Size and Duration Introduction Motivation vii

8 Literature Review Empirical Investigation: Preliminaries Empirical Investigation: Does Runup Size and Duration Change Over Time?. 148 Empirical Investigation: Autocorrelations of Runup Duration and Size Empirical Investigation: Is Runup Duration Correlated with Runup Size? Empirical Investigation: What Factors Affect the Size and Duration of Runup?..153 Data and Methodology Results 172 Implications Conclusion Appendix A Appendix B Appendix C References viii

9 TABLES Table 1. Model Summary Table 2. T-tests. 46 Table 3. T-tests. 47 Table 4. Weighted Least-Squares Regressions. 48 Table 5. Weighted Least-Squares Regressions. 50 Table 6. Variance Inflation Factors..52 Table 7. Weighted Least-Squares Regressions. 54 Table 8. Weighted Least-Squares Regressions. 57 Table 9. Regressions Table 10. Regressions Table 11. Univariate Results..124 Table 12. Ordinary Least-Squares Regressions.127 Table 13. Ordinary Least-Squares Regressions.130 Table 14. Ordinary Least-Squares Regressions.131 Table 15. Cumulative Abnormal Returns..132 Table 16. Coefficient Estimates.132 Table 17. Macroeconomic Factor Model Table 18. Target Runup Duration and Size 174 Table 19. T-tests.176 Table 20. Correlations 177 Table 21. Autoregression Models..178 Table 22. Macroeconomic Factor Model Estimations ix

10 Table 23. Ordinary Least-Squares Regressions Table 24. Ordinary Least-Squares Regressions x

11 FIGURES Figure 1. Plots of CAR(-50, t) xi

12 ESSAY I CHARACTERISTICS OF RUNUP BEFORE FULL ACQUISITIONS I. Introduction Acquisitions are one of the most salient events in the lives of firms. They are featured prominently in both the popular and business news. The popular press is interested for many reasons ranging from the effect on employment in the target s home town to the nationality of the acquirer. The latter business press, however, is interested because of the very clear reason that an acquisition implies a great deal of money changing hands. In this essay, I am concerned with determining when, and to whom, much of that money is transferred. Acquirers must pay an amount for each target share that is significantly (usually at least 10%) higher than the target s recent share price. Accordingly on the announcement of the acquisition the target s shares rise to reflect the offer for all of the target s shares at a price higher than the previous market price. If the acquisition is successful the target s shareholders are paid this, or a revised higher, price. 1

13 The prospect of such a large percentage gain in the targets shares in a short period attracts speculators 1. The speculators goal is to buy shares in the target before the announcement of the acquisition and profit from the subsequent increase in the target s share price. Some speculators seem to be successful in this endeavor, as evidenced in a price runup in the targets shares before the announcement. The price runup refers to consistent positive abnormal returns in the target s shares over a period immediately preceding the announcement of the acquisition. The goal of this paper is to ascertain particularly how these successful speculators gain their knowledge of the imminent acquisition. Further, I seek to find out how much these speculators know about the details of the acquisitions. I will also investigate what factors affect the timing of the speculators information acquisition. The most accepted broad explanation for the manner in which the speculators gain their knowledge is through leaked inside information. Herein I am interested in identifying the source of the leak, aspects of the acquirer and target which intensify the leak, and what information is contained in the leak. II. Contribution Previous papers have found factors that affect the size of target gains around the announcement of the tender offer. Particularly, Song, Stulz, and Walkling (1990) related the distribution of target ownership to the size of announcement gains. I will contribute to the literature by determining if the distribution of target ownership can affect target 1 The term speculator is used consistent with the definition of Harrison and Kreps (1978): An investor who is willing to pay more for a stock, given the right to sell the stock thereafter, than she would pay if constrained to hold the stock forever. 2

14 runup. If, in fact, much of the effect of the distribution of target ownership on target returns occurred in the runup period, then the results of Song, Stulz, and Walkling (1990) were likely understated. This may be the cause of the insignificance of some of their hypotheses. This paper seeks to contribute to the literature on informed trading. The manner in which an informed trader will attempt to profit from her information, and thus how the information the trader holds becomes incorporated into the market, was established by Kyle (1985). Continued work in the area of informed trading has found equilibrium security prices and trading profits when investors receive differing signals as to the security s ultimate value (Bernhardt and Miao (2004)). Herein I intend to add to the literature by investigating the channels by which information is leaked to the thereby informed trader. III. Literature Review Early work on leakage of inside information before merger and acquisition announcements, and the price runup that this leakage causes, was conducted by Keown and Pinkerton (1981). In their analysis they study 194 successfully acquired firms in the period The authors found that there was a significant leakage of information up to 12 days prior to the announcement of the bid. The size of the runup which information leakage causes has been analyzed extensively. Barclay and Warner (1993) documented that about 50% of the total premium received by a target in a full acquisition occurs before the announcement of the acquisition. In other analyses the percent of the total premium attributable to runup 3

15 ranges from 26% in Gupta and Misra (1989), to 43% in Keown and Pinkerton (1981), to 57% in Schwert (1996). In the analysis by Gupta and Misra (1988) they investigate whether insider trading is the cause of runup. Specifically, they intended to discern whether runup is caused by inside information or rather information which is publicly available. The authors found that targets, about which there were recent news stories, had significantly greater runup than targets which were not in the news. Much of the outstanding literature with respect to target stock runup before merger and acquisition announcements deals with runup as it relates to markup and the total premium paid for the target. An important paper in this regard is Schwert (1996) in which the author found evidence in favor of markup pricing whereby a dollar increase in the runup increases the total premium paid for the target by one dollar. This was evidence against the substitution hypothesis which proposed that an increase in the runup would be offset by a decrease in the markup, and thus the total premium the bidder paid for the target would be unaffected by the size of the runup. A result of Schwert (1996) was the important implication that any runup is a cost to the bidder. Schwert (1996) finds evidence against the substitution of runup for markup, and thus in favor of markup pricing, in a sample of completed acquisitions. That is to say, markup was defined as the difference between the final acquisition price (total acquisition premium) and runup. Betton, Eckbo and Thorburn (2008) find evidence against the substitution of runup for markup when defining markup as the difference between the initial offer price and runup. This definition of markup is unaffected by events which 4

16 occur after the initial bid and before the closing of the acquisition, such as further bidding. They find a dollar increase in runup increases the initial offer by $0.80. So, only 20% of runup substitutes for markup. Another strand of research has specifically investigated the effect investment banks advisors have on information leakage in mergers and acquisitions. This research was an offshoot of a comprehensive body of research with respect to the general role of investment banks in mergers and acquisitions. Servaes and Zenner (1996) found that target stock price runup before an acquisition announcement was not affected by whether the bidder used an investment bank to assist in the acquisition. In a present working paper Bodnaruk, Massa and Simonov (2008) find that if an investment bank advisor to the bidder holds a stake in the target, then the target s premium (the sum of runup and markup) increases to 36.5% from 30.6% in the case where the bidder s investment bank advisor does not hold a stake in the target. Further, the authors find evidence that if the bidder s investment bank advisor holds a stake in the target, then the probability that the bid fails is reduced. When the bidder s advisor holds a stake in the target, then the target tends to be overvalued and the post-merger operating performance of the combined firm is lower. Lastly, the authors find that less experienced bidders are more likely to take part in overvalued advisory induced deals. Recent analyses have found evidence that information leaks to the equity market about a firm through banks with which the firm has relationships. Acharya and Johnson (2007a) find information is revealed in the market for credit default swaps, which is later reflected in equity markets, in cases of adverse credit news. The amount of information 5

17 leaked is positively correlated with the number of banking relationships the firm has. In a present working paper, Acharya and Johnson (2007b), using a sample of leveraged buyout bids (LBOs) find there is more information leaked to equity markets if there are more equity participants in the financing of the purchase. Similarly, more debt participants in the financing of the acquisition implies more information leaked to the credit markets. There is also evidence of information leakage and informed trading before events other than mergers and acquisitions. Christophe, Ferri, and Angel (2004) find evidence of unusual levels of short-selling before negative earnings announcements. They conclude that the pre-announcement short-selling is likely caused by traders with information specific to the upcoming earnings announcement of the particular firm. An interesting result of the present analysis would be if the channels of information leakage which significantly affect target runup are common between mergers and acquisitions and earnings announcements. For example, it may be that the amount information leakage before both earnings announcements and acquisitions is affected by the amount of institutional ownership (of the firm announcing earnings and the target/bidder respectively). If in the present analysis institutional ownership is found to be a significant determinant of information leakage before merger and acquisition announcements, this may prompt a similar analysis regarding earnings announcements. Alternatively, if the channels of information, found in this analysis, are not relevant to earnings announcements, then the manner by which information is leaked may be significantly different in the cases of acquisitions and earnings announcements. 6

18 IV. Empirical Investigation of Factors Affecting the Magnitude of Runup A. Dependent Variable: Runup Event Windows A standard runup event window of (-42,-1), with t = 0 being the date of the bid announcement, was defined in Schwert (1996). To choose the event window, Schwert created a plot of the cumulative average abnormal returns (CARs) over the window (-126,253) for his entire sample of 1,814 mergers and tender offers. The plot showed that the CARs started to rise at approximately t = -42, and thus this day was set as the start of the runup event window. Notably, the sample of merger and acquisitions used in Schwert (1996) ranged from It is possible the date at which CARs become positive has changed in the years since This may be due to more recent bidders having a shorter or lengthier planning process before the initial bid. Similarly, the time required to secure the financing may have changed. As these time requirements change, so does the duration of time during which information may be leaked change. Accordingly, I will ensure that my results are robust to changes in the event window by calculating target runup over event windows both longer and shorter than the (-42, -1) event window of Schwert (1996). The Measure of Runup Runup is defined in the literature as the sum of the target s abnormal return for some event window before the announcement of the acquisition or tender offer (Schwert (1996)). Therefore to calculate runup for a target firm i, we must first estimate: (1) 7

19 where R it is the day t return for target i, R mt is the return on the market portfolio on day t, and t ranges over some period prior to the event window. We may then calculate runup as: (2) where (-a,-b) is the event window relative to t = 0 being the announcement date. Note the Latin characters are used here to denote estimated values. The Greek letters above are reserved for random variables. B. Explanatory Variables 1. Is target runup affected by whether the bidder specializes in acquisitions? The bidder has an interest in keeping runup as low as possible. If runup is not a perfect substitute for markup, then each dollar in runup costs the bidder. Information leakage about the impending bid for the target would cause runup, and therefore would reduce bidder returns from the acquisition. It follows then that the bidder would like to stop as much information leakage about the acquisition as possible. I hypothesize that firms which specialize in acquisitions such as Private Equity (PE) firms and Special Purpose Acquisition Corporations (SPACs) should be particularly adept in mitigating information leakage. Thus, I will test if acquisitions by such firms exhibit less runup. If these firms are successful in reducing information leakage, it may be because in the future they will be a continued source of revenue to the takeover industry. Any takeover adviser that leaked inside information about the current acquisition would risk being caught by the acquiring firm and losing future business. This would likely offset 8

20 any gains, in trading or otherwise, that the adviser would reap from leaking the information about the takeover. Hypotheses 1.1 H 0 : There is no difference in the size of the runup if the acquirer is a PE firm or SPAC. H a : Runup will be significantly smaller if the acquirer is a PE firm or SPAC. 2. Is target runup affected by the type of security issued to finance the deal? Often the acquirer will have to issue debt, equity, or other type of security to purchase the target. If this is the case then the board of directors of the target must consider the ability of the acquirer to issue the securities successfully. If the secirities issuance is unlikely to be successful, then the bid loses merit. Accordingly, the acquirer retains an investment bank responsible for placing the securities before submitting the bid to the board of directors of the target. Thus, by the time the bid is submitted, much work has been done regarding the placement of the securities. This opens another avenue for information leakage. Acharya and Johnson (2007a) found evidence that banks which lend to firms, and thus have inside information about the firm s financial condition, will hedge their exposure to their loan in the credit default swap market in the case of adverse news about 9

21 the borrower s financial condition. Moreover, the act of hedging in the credit default swap markets leaks information into the equity market, and causes a negative and significant correlation between credit default swap returns and future stock returns. This highlights information may be leaked through hedging in the debt markets, and not from trading strategies meant to profit from the inside information. However, recently in investigating information leakage with respect to leveraged buyouts, Acharya and Johnson (2007b) found that more debt and equity participants in a deal s financing implies more information leaked to the debt and equity markets respectively. This information leakage cannot be wholly explained through hedging activity. The outstanding literature finds evidence that information may be leaked directly by banks who assist in debt financing, for either hedging or for trading in the bank s own account. Moreover, information may be leaked unavoidably in the bookbuilding process. Thus, issuing securities to assist in the purchase of the target leads to the following encompassing hypothesis. The types of securities I will test for are financing by issuance of general borrowings, bridge loans, common stock, debt securities, loans from foreign lenders, junk bonds, lines of credit, mezzanine credit 2, rights offerings, and preferred stock. Including the above variables in a statistical analysis meant to explain the variation in target runup must be handled with care. An acquirer who must resort to mezzanine credit or common stock financing likely first exhausted all cheaper forms of 2 The term Mezzanine credit can be ambiguous. Here I mean issuances of unsecured and subordinated debt or preferred stock senior only to common shares. 10

22 financing (secured debt, bank loans, etc.). In the empirical analysis that follows only subsets of these variables will be included in any one regression equation. Hypothesis 1.2 H 0 : The acquirer issuing securities to assist in the purchase of the target has no effect on runup. H a : Runup will be greater if the acquirer issues securities to assist in the purchase of the target. 3. Is target runup affected by the payment type? The acquirer has three general choices for the method of payment in acquiring the target, or a mixture thereof. The three choices are cash, fixed dollar amount of bidder common stock, or a fixed exchange ratio of bidder versus target stock. The choice of the method of payment determines how the preclosing market risk and post-acquisition synergy uncertainty are divided between the acquirer and target. Preclosing market risk is the risk that the market s opinion, favorable or unfavorable, will be reflected in the share price of the acquirer as soon as the deal is announced. Post-acquisition synergy uncertainty is the risk that the proposed synergies will not be gained once the firms are combined. As the titles imply, preclosing market risk occurs between when the deal is announced and closed, and post-acquisition synergy uncertainty affects the combined firm after the deal closes. 11

23 In a cash purchase, the acquirer offers to pay a fixed amount in cash per share for the target firm. In this case the acquirer assumes all of the preclosing market risk and post-acquisition synergy uncertainty. The target assumes none of either risk. In a stock with fixed value payment, the acquirer specifies the value that each target share will receive in the acquirer s stock. This leaves the number of shares the acquirer pays for each target share fluctuating until the deal is closed. If the acquirer s share price rises the acquirer must pay fewer shares for each target share, and conversely if the acquirer s share price falls the acquirer must pay a larger number of shares for each target share. On closing, the acquirer will pay however many shares are required for the target s shareholders to receive the stated dollar price per share. In the fixed value stock payment, the proportional ownership of the merged firm is left uncertain until closing. The lower the value of the acquirer s shares at closing, the more of the combined firm the target s shareholders will own. This means that the postacquisition synergy uncertainty is split between the acquirer and target depending on the percentage of the combined firm each owns. Since the target shareholders are certain on the value they will receive for each share, the preclosing market risk is entirely borne by the acquirer. In the announcement of a fixed number of shares stock payment, the acquirer offers to give the target a certain set number of shares for each target share regardless of the value of the acquirer s shares at the close of the deal. In this case the target s shareholders are exposed to preclosing market risk in that if the acquirer s share price falls, the price paid for the target decreases. 12

24 Like the fixed value case, the fixed number of shares payment method leaves the post-acquisition synergy uncertainty split between the acquirer and target depending on the percentage of the combined firm each owns. However, in the fixed number of shares method, the proportional ownership of the combined firm is set from the announcement date of the deal. In each of these three cases the target shareholders take on a different amount of risk ranging from the least amount in the all cash case, to the most risk in the fixed number of shares of stock payment. It follows that the difference between the announced takeover price and the price at which the target s shares trade after the announcement should be least for the all cash payment, larger for the fixed value of stock payment, and largest for the fixed number of shares payment. This should hold true as investors are risk averse and require a greater expected return as risk increases. If the information leakage before the announcement of a takeover is of the form of a general rumor about an impending acquisition, then runup should be the same regardless of the method of payment that will be used. However, if the information being leaked is specific enough to include the method of payment, then we would expect runup to be greatest in the all cash payment case, less if the payment is a mixture of cash and stock, and least in the all stock case (encompassing both the fixed value and fixed number of shares method of payment). Thus, if runup is affected by the method of payment, then this tells us of the specificity of the information being leaked. The level of specificity narrows the sources 13

25 of the leak. If the method of payment is being leaked, then the information reaching the market must be from the bidder or an advisor to the bidder. Hypothesis 1.3 H 0 : The method of payment does not affect runup. H a : Runup is an increasing function of the percentage of the deal value which is paid in cash. 4. Is target runup smaller in hostile bids? In a hostile bid the target s management is less likely to know that a bid for the firm is imminent. Accordingly, there is a smaller probability that someone who has access to inside information from the target s management will purchase the target s stock. This leaves the most likely sources of inside information regarding the imminent bid as the bidder or someone working closely with the bidder. Since the bidder is motivated to keep information on the forthcoming bid from leaking, and the target s management cannot leak the bid for lack of knowledge, we would expect that the runup would be smaller for hostile bids than for friendly bids. In other words, comparing runup in hostile versus friendly bids effectively controls for any information being leaked by the target s management. With respect to this hypothesis, however, one must consider the results of Schwert (2000) concerning the lack of empirical differentiation between hostile and friendly bids. 14

26 In the analysis Schwert found that friendly and hostile bids are indistinguishable when using stock and accounting performance data to discriminate. Fundamentally, Schwert s analysis highlighted that all bids have both friendly and hostile aspects, and moreover a particular bid s classification as either hostile or friendly may vacillate before being ultimately defined. Hypothesis 1.4 H 0 : Target Runup is not affected by whether the bid is hostile or friendly. H a : Target Runup is smaller when the bid is hostile. target? 5. Is target runup affected by the percent of managerial ownership of the In the analysis by Song, Stulz, and Walkling (1990) the authors find that in the case where there are multiple bidders for the target, then the target s announcement gain increases with target managerial ownership. This may be because management requires that the takeover premium be no smaller than the rents management receives from the firm. Again no direct implication can be drawn from this result with respect to target runup. However, using the same argument as in the case of institutional ownership, it is logical that an informed trader would bid the target higher in the runup period if she expected the announcement gains to be higher due to the level of managerial ownership. Hypothesis

27 H 0 : There is no relationship between the proportion of the target s stock held by management and target runup. H a : There is a positive correlation between the proportion of the target s stock held by management and the target runup. 6. Is target runup affected by the percentage of target institutional ownership As noted in Song, Stulz, and Walkling (1990) institutional investors are usually in lower tax brackets, and thus they hypothesized that the target s gain decreases with higher institutional ownership. In their analysis the authors measured target gains as cumulative abnormal returns in a few days centered on the announcement date of the bid. Accordingly target gains in their analysis are more aptly called target announcement gains and any conclusion drawn from them is not immediately relatable to inferences on target runup as earlier defined in this analysis. Nonetheless, they found evidence supporting their hypothesis in a sample of 209 tender offers over the period 1968 through Due to such reasoning, runup may be inversely related to target institutional ownership. An informed trader may not be willing to bid the target s shares as high if there is a large proportion of institutional ownership in the target as the informed trader knows the institutional shareholders will not require large announcement gains in the target. The informed trader knows there is a possibility that either his information is incorrect, or that new developments may cause the bidder to not actually make the bid. If 16

28 no bid is made the informed trader may well lose if the target falls to its pre-runup price. Accordingly the informed trader weighs the benefits in announcement gains against the loss from the bid not being made when deciding how high to bid the target s shares in the runup. The lower the benefit in announcement gains, then to lower heights is the informed trader willing to bid the target s shares. Hypothesis 1.6 H 0 : There is no relationship between the level of institutional ownership in the target and target runup. H a : There is an inverse relationship between the level of institutional ownership in the target and target runup. 7. Is target runup affected by the proportion of the target that the bidder owns at the time of the bid announcement? Song, Stulz, and Walkling (1990) found that a larger proportion of the target owned by the bidder before the initial takeover offer tends to decrease the target s takeover gain. Their reasoning was that the more of the target s shares the bidder owns, the lower on the supply curve of the target s shares is the marginal shareholder who must tender shares for the offer to succeed. This lower point on the supply curve implies the bidder may buy a controlling stake in the target at less cost. Again this may not be directly applied to the case of runup, though by the same argument in the cases of 17

29 managerial and institutional ownership, we may expect an informed trader to not bid the target as high if she knew the bidder already had a large stake in the target. In a present working paper Betton, Eckbo, and Thorburn (2008) find that if a bidder, which will ultimately make a control bid for the target, purchases a short-term toehold before the control bid then this increases target runup. A short term-toehold is defined as a toehold purchase in the runup event window of (-42, -1) with respect to a control bid at t = 0. This excludes toeholds the bidder may own in the target so long as that toeholds was purchase before t = -42 relative to the control bid. In my analysis, I will consider both short and long term toeholds. That is to say, I will not exclude toeholds which were bought prior to the runup event window in the control bid. This analysis will therefore include the effect of a longstanding relationship the bidder may have with the target predicated on or evidenced by the bidder s stake in the target. Unfortunately this methodology also may include negotiated staged takeovers, where the bidder and target agree that a full-takeover will follow the bidder s toehold contingent on financial goals being met or other events occurring. However, such contingencies on the full-takeover are present to some extent in most acquisitions. Hypothesis 1.7 H 0 : There is no relationship between the proportion of the target s stock held by bidder before the announcement and target runup. H a : There is a negative correlation between the proportion of the target s stock held by the bidder before the announcement and target runup. 18

30 acquired? 8. Is target runup affected by the ex-ante probability that the target is Song and Walkling (2000) created an acquisition probability hypothesis which proposed that rival firms of acquisition targets earn abnormal returns at the announcement of the acquisition because the probability that they will become targets themselves increases. In their analysis the authors use a logit model to calculate the probability that a firm will become a takeover target. The authors then use a linear regression model to estimate the effect that these predicted probabilities have on the abnormal returns to rivals of acquisition targets. This regression finds that the rival abnormal return is significantly and positively related to the estimated probability that it will be an acquisition target. This is somewhat indirect evidence of the acquisition probability hypothesis, as the above test correlated rival abnormal returns with the level of probability and not the change in the level of probability as required in the hypothesis. Similarly, herein I will estimate each target s ex ante probability of being acquired using a binary response model. I will then test whether this estimated probability explains any variation in the size of the target s runup. Consistent with Song and Walkling (2000) I expect that the greater the probability of acquisition, the larger will be the target s runup. Hypothesis 1.8 H 0 : There is no relationship between the ex ante estimate of the probability that the target will be acquired, and the size of the target s runup. 19

31 H a : The larger the ex ante estimate of the probability that the target will be acquired, the larger will be the target s runup. Binary Response Model of the Target s Ex Ante Probability of Being Acquired I will calculate the probability of a firm becoming the target of an acquisition by using a binary generalized linear model. The dependent variable is 0 if the firm is not a target and 1 otherwise. I will use explanatory variables similar to those in Song and Walkling (2000) by including the firm s growth rate (3-year annualized sales growth), size (measured as the log of market capitalization), Tobin s Q 3, managerial ownership (percentage), institutional ownership (percentage), and leverage. Consistent with previous literature, and because the logistic scale is appropriate regardless of whether the data are sampled prospectively or retrospectively, I will use a logit link function in the binary generalized linear model. Denote the (nxp) matrix of explanatory variables as X, where there is a column p for each explanatory variable, and each column p contains the n observations across firms. Let β be the (px1) vector of parameters. The logit link function is then: (3) where the left hand side of the equation is referred to as the log-odds ratio. 3 I will use Chung and Pruitt s approximation for Tobin s Q: where MVE is market capitalization, PS is the liquidating value of preferred stock, DEBT is short-term liabilities less short-term assets plus the book value of long-term debt, and TA is total assets. These data are gathered from Standard and Poor s COMPUSTAT North America. 20

32 Before I apply the parameters of the logit model to estimate the ex ante probability that each target in my sample becomes a takeover target, I must estimate these parameters on a sample of firms which both do and do not become takeover targets. As can be seen in section VI-A, my sample of target firms begins on January 1, Thus, to estimate the parameters of the logit model I will choose firms which were targets prior to Specifically, I will choose a random sample of 50 firms which were targets of successful bids for control in the years I will choose the targets with the same constraints as I choose the targets in my main sample as illustrated in section VI-A. These targets will be selected from Thompson s SDC Mergers and Acquisitions database. I will calculate the mean, maximum, and minimum firm size for the sample of 50 targets. I will then choose a random sample of 50 firms from the population of firms which both have existed since 1998 and have not been the target of a successful control bid. This random sample will have the same characteristics as the sample in section VI- A, with the exception that they are not targets. Moreover, I will constrain the random sample of non-targets to have a mean firm size within one standard deviation of the mean firm size of the sample of 50 targets. The firms in the sample of non-targets must also have a firm size within the maximum and minimum firm size of the sample of 50 targets. For the logit model estimation I will gather the data on sales growth, size, Tobin s Q, and leverage from COMPUSTAT North America. Data on institutional and managerial ownership will be gathered from Share-World and proxy statements respectively. The data will be as of the end of the fiscal year preceding the second half of

33 (SOX)? Has target runup been affected by the introduction of Sarbanes-Oxley Recent research by Madura and Ngo (2008) has found that aspects of mergers have changed since the introduction of SOX. Among their conclusions are some which motivate an interest into how SOX may have affected runup. Particularly, the authors found that targets are more intensive in their use of outside advisors since the passage of SOX. Therefore, if information is leaked through outside advisors, then runup should increase post-sox. The authors also found that the total premium (runup plus markup) paid by the acquirer for the target has decreased significantly since SOX. Considering Schwert (1996) found that runup is uncorrelated with markup, it is possible that the lower total premium since SOX is due to both lower runup and markup. If so, then runup should decrease post-sox. Further, the authors found that since SOX the post-merger long-term stock price performance of the combined firm has improved. This result implies purchasing shares in the target during the runup period and holding those shares has been more profitable since the introduction of SOX. This may assuage the concern that a speculator trading on inside information may have about being able to sell out of a long position in an illiquid stock. This motivates the use of an interaction term between target liquidity and a post- SOX dummy. 22

34 The above reasoning with respect to the impact of the passage of SOX on target runup leads to two-tailed hypotheses. Hypothesis 1.9 H 0 : There is no relationship between the passage of SOX, and the size of the target s runup. H a : The passage of SOX has significantly affected the size of runup. Hypothesis 1.10 H 0 : There is no relationship between the interaction of the passage of SOX and target liquidity. H a : The interaction of the passage of SOX and target liquidity, significantly affects the size of the target s runup. 11. Is target runup affected by whether the acquirer is foreign? The extent to which a foreign firm can hide its acquisition intentions in the U.S. may differ from a domestic firm. Advisors to the foreign firm may feel more free to divulge inside information if they think the foreign acquirer does not have sufficient domestic legal representation. This may increase the target s runup. Alternatively, preparations the foreign acquirer may make for the acquisition, such as arranging financing, may not be an avenue through which information may be 23

35 leaked as in the case of a domestic acquirer. A foreign acquirer may use a financing source that does not have extensive operations in the U.S. In this case the target s runup would be smaller. This leads to the following two-tailed hypothesis. Hypothesis 1.11 H 0 : There is no relationship between whether the acquirer is foreign, and the size of the target s runup. H a : Whether the acquirer is foreign significantly affects the size of target runup. 12. Is target runup affected by whether there is management participation? Given the result of Schwert (1996), that implied there is little substitution between runup and markup, it is clear that runup is a cost to the acquirer. In the case of management participation in the deal, the acquirer is partially the target s own management. If information is leaked through the target s management, then in the case of management participation this information leakage should be reduced as it is a cost to the management. Hypothesis 1.12 H 0 : The size of runup is not affected by whether there is management participation in the acquisition. 24

36 H a : Runup is significantly smaller in the case of management participation in the deal. 13. Is target runup affected by the existence of listed options on the target s stock? If listed options are available to trade on a target s stock, then informed traders may choose to trade the option instead of the stock. Therefore, an informed trader with information about a target with listed options may profit from that information in either the option or stock market. The additional trading venue may increase demand for information relating to targets with listed options. Of stocks for which there are listed options, Jayaraman, Frye, and Sabherwal (2001) find that there is a significant increase in trading of both call and put options in firms involved with a takeover, prior to the announcement of the takeover. They also find that a significant amount of the options trading is being done by informed traders. Further, Jayaraman, Frye, and Sabherwal (2001) find evidence that the trading in the options market leads the trading in the underlying stock. Ni, Pan, and Poteshman (2008) find that there is informed trading on stock volatility in the options market before earnings announcements. Brent, Morse, and Stice (1990), Chen and Singal (2003), and Senchak and Starks (1993), all show that the level of short interest in a stock is affected by whether there are listed options available to trade on that stock. Hypothesis

37 H 0 : Runup is not affected by the existence of listed options on that stock. H a : Target runup is greater if there are listed options available to trade on that stock. acquirer? 14: Is target runup affected by the size of the target relative to the size of the The higher the ratio of the target s size to the acquirer s size, the larger the target relative to the acquirer. A higher ratio implies that for the acquirer to be successful there must be more cooperation on the part of the target. This leaves that target in a better position to demand information from the acquirer before agreeing to the merger. Since target runup is a cost to the bidder and a gain to the target, the target has an economic interest in leaking such information increasing target runup. Hypothesis 1.14 H 0 : Target runup is not affected by the size of the target relative to the acquirer. H a : Target runup is positively correlated with the size of the target relative to the size of the acquirer. 15: Is target runup affected by hedge fund involvement in the deal? 26

38 In general, the more participants involved in a merger or acquisition the more numerous the avenues for information leakage. Therefore it follows if hedge funds are additionally involved in any aspect of the deal (such as in purchasing a stake in the target and voting for the acquisition, or in the deal s financing, etc.), then information leakage should be greater which will cause greater target runup. Moreover, hedge funds are known to speculate on possible acquisition targets, and so their involvement in the deal may be motivated by data gathering. Hypothesis 1.15 the deal. H 0 : Target runup is not affected by whether there is hedge fund involvement in H a : Target runup is greater when there is hedge fund involvement in the deal. 16: Is target runup affected by the acquirer being a financial firm? Financial firms are regular participants in the capital markets. As such, their borrowing for the acquisition may be more easily hidden than in the case of a firm which rarely borrows. Moreover, financial firms should have more knowledge about capital market operation than the average bidder, and similarly may be more able to arrange financing undetected. This will reduce the amount of information leakage. Hypothesis 1.16 H 0 : Target runup is not affected by whether the acquirer is a financial firm. 27

39 H a : Target runup is smaller when the acquirer is a financial firm. advisors? 17: Is target runup affected by the number of acquirer financial and legal The acquirer has an economic incentive to minimize information leakage as Schwert (1996) found that target runup and markup are uncorrelated, implying target runup is a cost to the bidder. The more financial and legal advisors an acquirer has, however, the more likely it is that either information is unintentionally leaked or that an advisor uses the information to profit. So I expect that target runup will be an increasing function of the number of acquirer financial and legal advisors. Hypothesis 1.17 advisors. H 0 : Target runup is not affected by the number of acquirer financial and legal advisors. H a : Target runup increases with the number of acquirer financial and legal 18: Is target runup affected by the amount that the acquirer spends on financial advice? The amount that the acquirer spends on financial advice is measured as the acquirer financial advisor fees as a percentage of the deal value. I expect that the more 28

40 the acquirer spends on advice the better advice the acquirer will receive. Such advice may be regarding how to lessen information leakage as it is a cost to the bidder. Further, a greater amount spent by the bidder on advice may be a signal that the bidder is very interested in completing the deal. If the bidder wants to offer a high markup, to make the success of the offer more likely, then the bidder should minimize runup. Hypothesis 1.18 advice. H 0 : Target runup is not affected by the amount the acquirer spends on financial H a : Target runup is negatively correlated with the amount the acquirer spends on financial advice. fee? 19: Is target runup affected by the structure of the acquirer financial advisory If the acquirer financial advisory fee is a flat-rate, then the advisor may be more likely to seek to profit on inside information once the advisory fee is earned. In such fashion a flat rate advisory fee may increase target runup. Conversely, a flat rate advisory fee may reduce the amount of time that the advisor works on behalf of the acquirer and thereby shorten the window in which information may be unintentionally leaked by the advisor. Thus, a flat rate advisory fee may conceivably increase or decrease target runup leading to a two-tailed hypothesis. 29

41 Hypothesis 1.19 a flat rate. H 0 : Target runup is not affected by whether the acquirer financial advisory fee is H a : Target runup is affected by whether the acquirer financial advisory fee is a flat rate. advisors? 20: Is target runup affected by the number of target financial and legal More target financial and legal advisors implies more avenues of information leakage. Thus, all else constant, I expect target stock price runup to be an increasing function of the number of target financial and legal advisors. Moreover, I expect the target s financial advisors to be familiar with the result of Schwert (1996) that target runup is uncorrelated with markup and therefore target runup is a cost to the bidder and a gain to the target. It follows that the presence of financial advisors may cause the target to be more likely to leak information about the imminent bid, or to tolerate such leaked information. Hypothesis 1.20 advisors. H 0 : Target runup is not affected by the number of target financial and legal 30

42 advisors. H a : Target runup is greater when there are more target financial and legal 21: Is target runup affected by whether the deal is a tender offer? In a tender offer the acquirer offers to buy shares from the owners generally bypassing target management. The less dealing the acquirer has with the target management, the less target management will know and be able to leak. This is important as target management have an economic incentive to cause target stock to runup before an acquisition announcement if runup and markup are uncorrelated as found in Schwert (1996). Thus in the case of a tender offer, target runup should be reduced. Schwert (2000) showed a hostile acquisition is more likely to be a tender offer. Further, many acquisitions deemed hostile were started in a friendly manner, and were only classified as hostile after negotiations between target management and the acquirer broke down prompting the acquirer to launch a tender offer. I will therefore estimate the parameters of regression models which contain only one of the tender offer and hostile deal variables.. Hypothesis 1.21 H 0 : Target runup is not affected by whether the deal is a tender offer. H a : Target runup is smaller when the deal is a tender offer. 31

43 22. Is target runup affected by the target s size? If a target has a larger market capitalization, then there is a greater opportunity to profit from inside information. It follows that inside information with respect to larger targets would be in greater demand. This may cause the inside information to be disseminated more widely than in the case of a larger target, thereby increasing runup. However, if there is a point where the target s size no longer affects information leakage and those with the inside information have limited capital, then a larger target should experience smaller runup. Moreover, this may still hold even if those with inside information do not have limited capital in that devoting a too large percentage of their investments to the target may signal insider trading to market regulators. According to Roll (1984), there is a strong inverse correlation between firm size and stock liquidity (as measured by the bid-ask spread). Thus, target firm size will also control for target stock price liquidity. The liquidity of the target s stock may be an important factor in whether an informed trader can buy shares in the target unnoticed. Muelbroek (1992) found that on days when insiders trade the target s stock during the runup period, total volume traded in the stock is greater than expected. This is evidence that the abnormal returns in the runup period are due to the increased insider trading volume. Further, Muelbroek (1992) finds that the target s stock price is affected by trade specific characteristics such as the insider s trade direction, trade size, number of trades, as well as the total volume traded by the insider. Therefore, I hypothesize that the less liquidity in a target s stock, the more readily the market can identify insider trading which will cause greater runup. 32