Jurisdictional Effects in M&A Litigation

Transcription

1 Jurisdictional Effects in M&A Litigation C.N.V. Krishnan, Ronald W. Masulis, Randall S. Thomas, and Robert B. Thompson * We compile the most extensive hand collected data set on all forms of M&A litigation in the U.S. to study the effects of lawsuit jurisdictions during a sample period (1999 and 2000) of the Fifth Merger Wave, a period characterized by an abundance of friendly one-bidder deals and the near demise of the hostile offer. We find that only about 12% of all M&A offers are challenged in the courts during this period. Almost half of the suits are filed in Delaware, while federal suits account for less than 9% of the suits in our sample. We find a very small incidence of multi-jurisdictional litigation (about 3% of all suits), unlike the recent sharp increase in such cases in the post-2008 global financial crisis period. We find that litigation filed in Delaware is less of a barrier to deal completion than cases brought in federal court. Litigation filed in federal court is associated with a significantly higher takeover premium in all offers and in completed deals, suggesting that state court cases, on average, put less pressure on bidders to raise takeover premia. In line with these findings, we find that federal courts attract a significantly higher proportion of target initiated litigation than state courts; no target lawsuit is filed in Delaware during our sample period. Finally, we find that while jurisdiction does not significantly affect settlement rates or the consideration paid upon settlements, litigation challenging controlling shareholder squeeze-outs is more likely to settle with cash consideration paid to shareholders, reflecting the stricter judicial standard applied to such bids. * Address correspondence to Randall Thomas at randall.thomas@vanderbilt.edu. C.N.V. Krishnan is at Weatherhead School of Management, Case Western Reserve University, cnk2@case.edu; Ronald W. Masulis is at Australian School of Business, University of New South Wales, ron.masulis@unsw.edu.au; Randall S. Thomas is at Vanderbilt University Law School; and Robert B. Thompson is at Georgetown University Law Center, thompson@law.georgetown.edu. 1

2 I. INTRODUCTION Using the most extensive hand collected data set of Mergers and Acquisitions (M&A) litigation in the U.S. currently available, this study documents the effects of lawsuit jurisdictions on deal outcomes and lawsuit settlement. The analysis is based on a sample of 2512 distinct M&A offers by U.S. acquirers announced during the period, which, as shown in Krishnan, Masulis, Thomas and Thompson (2012), is a representative period for studying M&A litigation in the Fifth Merger Wave period ( ) that is dominated by friendly singlebidder offers and few competitive or hostile offers (Andrade, Mitchell and Stafford 2001). This period of analysis is preceded by the Fourth Merger Wave ( ), known for its prominent hostile transactions and frequent bidder and target firm suits (Gilson and Black 2004), and followed by the Sixth merger wave ( ) where cash financed transactions are very common (Alexandridis, Mavrovitis and Travlos 2011). The Sixth merger wave was followed by the post-2008 global financial crisis period where we see a rising percentage of M&A deals facing lawsuits and multi-jurisdictional litigation (Daines and Koumrian 2012; Cain and Davidoff 2012). Using a much more reliable dataset, our study provides a clear baseline for comparison with other periods of M&A activity. Our data show that the most common form of M&A suits is shareholder class actions (representing about 86% of all cases) which typically allege that target company directors breached their fiduciary duties to their shareholders by selling the company too cheaply. In contrast to the Fourth Merger Wave, we find very few bidder company suits claiming that a target board of directors who refuses to sell the company is in breach of its fiduciary duties to shareholders, and very few target company suits alleging that a bidder has violated federal securities, antitrust law or state corporate law by making its offer. 1 Individual direct actions, 1 Jarrell (1985) reports that target managements litigate in roughly one third of all takeover attempts in the Fourth Merger Wave. 2

3 derivative suits, antitrust cases, and multi-jurisdictional litigation, are also all relatively uncommon. 2 In contrast to the post-2008 global financial crisis period, we find lawsuits challenge less than 12% of M&A deals in our sample. Some scholars claim that since the late 1990 s Delaware has continuously lost M&A litigation to other jurisdictions (Armour, Black and Cheffins 2012). Our data provide the first definite benchmark on what M&A litigation looks like at the beginning of this period. We find that slightly less than half of M&A lawsuits in our sample are filed in Delaware Chancery Court, less than nine percent are filed in federal courts, and roughly forty percent are filed in state courts other than Delaware. Also, unlike the dramatic rise in multijurisdictional litigation witnessed in the post-2008 global financial crisis period (Cain and Davidoff 2012; Daines and Koumrian 2012), we find comparatively few litigated M&A offers involve multi-jurisdictional litigation, i.e. 3% of all litigated offers, where lawsuits challenging a single transaction are filed in both federal and state courts, or in more than one state court. Using multivariate analysis, where we control for lawsuit type, transaction features and industry, year and offer-type fixed effects, as well as for selection bias, we find that deals litigated in Delaware are more likely to be completed compared to cases brought in federal court. However, takeover premia in deals subject to state court litigation are significantly lower than deals subject to federal court litigation. These findings suggest that in deals subject to federal court litigation, fewer bids are completed, but at significantly higher premiums compared to deals subject to state court litigation. Indeed, federal courts attract a significantly higher proportion of bidder and target initiated litigation, which are correlated with lower deal completion rates, but higher takeover premiums in completed deals. In fact, no target lawsuits are filed in Delaware during our sample period. We also find that multi-jurisdictional litigation results in a significantly lower deal completion rate. 2 There were only 2 antitrust cases, both of which entailed piggyback private actions that followed the announcement of a government enforcement case. 3

4 Finally, examining M&A lawsuit settlements, we find that only one third of the cases filed are settled by the contesting parties. When we decompose these settlements, we see that 42% entail higher deal premiums paid to shareholders (cash consideration), while an additional 18% receive other substantive relief. In comparison, the M&A litigation settlement rate is much higher in the subsequent period reaching almost 72% (Cain and Davidoff 2012), while the cash consideration payment rate is much lower at 5% in the post-global financial crisis period (Daines and Koumrian 2012), although these two studies only examine large offers. Interestingly, lawsuit settlement rates are not significantly different for Delaware and federal court cases. The consideration paid in lawsuit settlements is significantly higher in Delaware relative to other state courts, but only in univariate tests; after controlling for lawsuit types (especially whether the lawsuit pertains to controlling shareholder squeeze outs), this effect disappears. Litigation pertaining to controlling shareholder squeeze-outs, which are most often filed in Delaware court, have higher settlement rates and are more likely to yield cash awards, which highlights the more plaintiff-friendly legal regime that prevails in controlling shareholder squeeze-out cases. About two-thirds of multi-jurisdictional cases that settle involve payments of monetary awards, more than other lawsuits. These cases challenge significantly larger M&A transactions relative to other lawsuits. This pattern is consistent with potentially more lucrative cases attracting more litigation in a wider set of legal forums. The remainder of the paper is organized as follows. The next section reviews related prior literature, Section III details our sample selection procedure and describes our data, Section IV conducts extensive univariate and multivariate analyses of the associations between lawsuit jurisdictions and outcomes, and Section V concludes. The Appendix reports on additional robustness analysis for our main results, and provides definitions of key variables used in our analysis. 4

5 II. RELATED LITERATURE The objective of this paper is threefold: (a) to analyze the effects of lawsuit jurisdiction on deal and lawsuit outcomes, (b) to examine the determinants of the probability of lawsuit settlements and consideration paid upon lawsuit settlements, and (c) to provide the first definite benchmark for what M&A litigation looks like before the well-documented findings on litigation by jurisdiction in more recent times. The paper most closely related to the current study is Krishnan, Masulis, Thomas, and Thompson (2012), which examines one form of M&A litigation, namely shareholder class action lawsuits, and their effects on deal completion probability and takeover premium, over the period. For shareholder class actions, it finds that the expected rise in takeover premia more than offsets the fall in the probability of deal completion, resulting in expected gain for target shareholders. That study concludes that during the Fifth Merger Wave, characterized by friendly single-bidder offers, shareholder litigation substitutes for the existence of a rival bidder by challenging low-ball bids, often forcing offer price improvement. While we use the same sample period as in Krishnan, Masulis, Thomas, and Thompson (2012), our focus is different. We present analysis of M&A litigation within the broader context of examining jurisdictional - Delaware, federal, non-delaware state, and multijurisdictional - effects on deal and lawsuit outcomes for a wider range of M&A litigation classes. In Krishnan, Masulis, Thomas, and Thompson s (2012) examination of shareholder litigation, the only jurisdictional variable is for Delaware suits, which was included as a control variable. In contrast, in the current study, three jurisdictional control variables are included in our analysis: namely, Delaware suits, Federal suits and multijurisdictional suits. (Note that a litigated offer can involve a suit in Delaware court, Federal court, a state court other than Delaware, or be filed in multiple jurisdictions, which are all variables of interest in this study). Krishnan et al. s (2012) major jurisdictional finding is that Delaware is positively, but insignificantly related to deal completion success (t-statistics of 1.36 in one specification), and 5

6 insignificantly negatively related to takeover premium. However, this prior study does not examine the effects of Federal suits and multijurisdictional suits. Moreover, our current study includes a broader range of litigation data than those studied by Krishnan, Masulis, Thomas, and Thompson (2012) (adding major M&A lawsuit classes other than just shareholder class action lawsuits), which also increases the overall proportion of Federal suits in the full sample. Thus, both the changes in the regression specification to focus on a wider range of legal jurisdictions and the expansion in the sample of additional classes on M&A lawsuits makes it possible for this study to clearly identify the effects of legal jurisdiction on M&A outcomes and settlement activity, which is not investigated in the previous study. The results in this study show that a Delaware filing is significantly positively related to deal completion at the 5% significance level, but is insignificantly related to takeover premium. Federal filings are significantly positively related to takeover premium in all offers at the 5% significance level and to takeover premium in completed deals at 10% significance level, but are insignificantly related to deal completion rates. Multijurisdictional litigation is significantly negatively related to deal completion success at the 10% significance level. We believe we are the first to document these findings. Also new, is our examination of lawsuit settlement rates and consideration paid upon lawsuit settlements. We report that lawsuit jurisdiction does not significantly affect settlement rates or the consideration paid upon settlements. However, lawsuits challenging controlling shareholder squeeze-outs have a significant positive correlation with monetary settlements, reflecting the strength of the underlying legal claims in these cases. Among the recent related papers, Daines and Koumrian (2012) look at shareholder litigation targeting large M&A deals (valued at over $500 million) and find that over 95% of such offers in are litigated, and of the settled cases, only 5% resulted in payments to shareholders and a large majority of these settlements (82%) only involve additional disclosures, which is a reversal of our finding that less than 12% of offers are litigated and 42% 6

7 of settlement cases include cash consideration in the previous Fifth Merger Wave period. Cain and Davidoff (2012) examine a sample of large U.S. M&A deals of over $100 million in value, announced in the period, which includes the 2008 global financial crisis as well as several years before and after. In comparison to the Fifth Merger Wave, they find a much larger percentage of offers (56%) attract litigation and that increasingly these cases are multijurisdictional in nature (increasing to 53% in 2011). In sum, in the post-global financial crisis period, there is a trend towards an increase in the percentage of M&A transactions attracting litigation and a sharp rise in multi-jurisdictional litigation. In comparison, our study includes a much broader set of deals, including smaller ones from the Fifth Merger Wave, which allows us to determine appropriate more general baseline for comparisons with these recent studies of large M&A offers. III. DATA A. Lawsuits We assembled our litigation dataset of mergers and acquisitions cases filed over a two year period (1999 and 2000) 3 using two approaches. For litigation occurring in Delaware state courts, the most important American jurisdiction for corporate law and lawsuits, we compiled all complaints and other relevant documents filed for every case brought in the Delaware Chancery Court, and identified the action or transaction that gave rise to the litigation. When the subject of the litigation was a merger or acquisition, as most are, we tracked the deal s progress and the litigation until it was resolved. Using this method, we uncovered all mergers and acquisitions litigation filed in the Delaware Chancery Court during 1999 and For litigation occurring outside of Delaware (where electronic filing adoption has occurred more slowly and where there are many more courts in which cases can be filed), we rely on companies federally mandated securities law filings to determine whether litigation 3 Krishnan et al. (2012) argue that this sample period is representative of the Fifth Merger Wave in the U.S., since they find no significant differences in terms of offer features and outcomes in these two years as compared to the full Fifth Merger Wave sample period ( ). 7

8 surrounding their merger and acquisition activities is filed. Our search procedure is to examine all target (and if necessary bidder) SEC filings (including Form 10-K, Form 10-Q, Form 8-K, Form S-1, Schedule 14A, Schedule 14D9, and Schedule TO) on and after the announcement date of the M&A transaction. If we do not find disclosures within two months of the litigation announcement date challenging the transaction, then we code the transaction as not generating litigation. This is consistent with earlier research reporting that almost all M&A litigation is filed within one month of a proposed deal s first public announcement (Thompson and Thomas 2004). This is supplemented with a search of the Lexis-Nexis database for press releases and other deal related announcements for evidence of deal litigation. Federal securities law (via Item 103 of Regulation S-K incorporated into the various forms described above) requires disclosure of litigation if it is material, a term that lacks a bright-line on-off switch and is capable of varying interpretations by different companies and their legal advisors. Thus, we cannot be sure that we have found all M&A litigation cases filed in courts outside of Delaware, only that we have found all material mergers and acquisitions cases (as interpreted by issuers and their lawyers) filed in those courts. We address this issue of potential bias in two different ways. First, we find that 12.3% of proposed deals involving Delaware corporations generate litigation, while 11.5% of proposed deals involving non-delaware corporations lead to litigation. This difference is not statistically significant. Given that we are confident of uncovering virtually all the Delaware M&A litigation, and that there is no a priori reason to believe that M&A deals involving Delaware corporations are more likely to generate litigation than deals involving non-delaware firms, we view this evidence as suggesting that any potential sample bias is of minor concern. Second, we re-estimate the coefficients of our main regression model using weighted least squares methodology, after assuming that we uncovered only 50% of the non-delaware suits, and find that our main results continue to hold. 8

9 A second challenge in our collection efforts is variation in the amount of information reported for each identified litigation case. Some companies disclosed not only summaries of the allegations of cases filed, but also copies of the actual complaints filed in the cases, while in other companies the level of disclosure is quite incomplete. To follow up on cases in the latter category, we search the federal electronic data system (PACER) and state electronic filing systems and supplement both of these data sources with searches of Lexis Court Link, which is a for-profit service that permits retrieval of dockets and certain other information for some federal and state cases. In addition, Vanderbilt Law School librarians contacted each court individually to request additional information concerning the case and its outcome. We also wrote letters to all the attorneys of record that we could identify in each case, asking them for missing pieces of information needed for our analysis. Mergers and acquisitions litigation is filed in different courts and in different forms. While the Delaware Chancery Court is recognized to be the most important location for merger and acquisition litigation, filings can occur in any federal district court in the country or in state court in any one of the other 49 states. Litigation over the same transaction may also be filed in multiple courts by different parties. One important issue that our data allows us to examine is whether Delaware is different from other states in how it treats merger and acquisition litigation. Given its preeminent position, commentators have argued over whether it provides a more favorable forum for plaintiffs or a less favorable one. A second important related issue is the extent to which multijurisdictional litigation is filed. For this reason, we track cases by the court in which they are filed. We break the sample into three broad categories: Delaware Chancery Court; Other State Courts, and Federal Court cases. It is important to note that mergers and acquisitions lawsuits can be filed as class actions, derivative lawsuits, individual actions by either the bidder or the target company or a shareholder, and a few other less common forms. Class actions and derivative lawsuits are 9

10 representative litigation in that a plaintiff's law firm chooses to pursue the matter, nominally at the request of a particular named shareholder(s), on behalf of the company itself or all its shareholders that are allegedly adversely affected by the target company's actions. Plaintiff lawyer incentives in representative litigation may be significantly different from those of lawyers hired by management to represent a bidder or target (Cox and Thomas 2009). Target management and bidders may also have different interests from that of target shareholders (Rosenzweig 1986). For these reasons, we need to control for each form of litigation as there could be significant differences in their impact on the transaction. We classify cases into the following categories: class action lawsuits, individual direct action lawsuits, derivative lawsuits, bidder lawsuits, target lawsuits and multiple lawsuits (offers in which more than one type of lawsuit was filed). Another important dimension of mergers and acquisitions litigation is whether or not it results in a direct benefit to target shareholders. Many cases are dismissed before reaching a judgment, either by the plaintiffs themselves or by the court at the behest of the defendants. However, a significant portion of cases filed result in either cash payments to target shareholders, or an equivalent benefit in the form of an increase in the takeover premium (Thompson and Thomas 2004). We are interested in determining whether the location of the court hearing the case influences lawsuit settlements. However, the type of transaction may influence the strength of shareholder legal claims and the price that they receive for their shares. For example, potential conflict of interest transactions, such as controlling shareholder squeezeouts, may result in lower premiums being offered to departing shareholders, but generate stronger legal claims for dissenting shareholders. By contrast, hostile transactions may result in higher premiums in completed deals, but fewer potential shareholder suits with cash settlements (Thompson and Thomas 2004). For these reasons, we control for transaction type in our analysis. 10

11 Our analysis should provide a more comprehensive and accurate picture of the jurisdictional effects of litigation than studies that rely solely on the SDC database for litigation information. 4 Indeed, comparing the Thomson Reuters SDC Platinum Mergers & Acquisitions database with our hand collected data, we find that the SDC database fails to report 10.3% of Delaware Court litigated M&A deals that we uncover in our initial dataset. Moreover, the SDC database also does not provide details on litigation types and jurisdictions or settlement outcomes that are needed for our analysis. B. Other Control Variables While determining the associations between litigation and M&A offer outcomes, we need to control for several important offer characteristics that prior research finds, influence M&A offer outcomes: (a) Intra-Industry Offer, an indicator variable for when both bidder and target firms are from the same (2-digit SIC code) industry, which serves as a proxy for less severe information asymmetry problems and greater expected synergies [see Andrade, Mitchell and Stafford (2001)]; (b) Offer Size, the value of the transaction (in $ billion), measured by total consideration paid excluding fees and expenses, which captures economic deal complexity (see Servaes and Zenner 1996; Moeller, Schlingemann and Stulz 2004); (c) Hostile Offer, an indicator variable equal to one for hostile bids and unsolicited offers, as reported in the SDC database, following Officer (2003), which measures deal completion difficulty (see Bebchuk, Coates, and Subramanian 2002); (d) Multiple-Bidder Offer, an indicator variable equal to one for offers involving competing bidders, which also measures deal completion difficulty (see Bradley, Desai and Kim 1988); (e) Stock Financing, the percentage of the total offer price paid in bidder or bidder subsidy stock, which captures market timing and other stock-price related issues (see Loughran and Vijh 1997), or alternatively, Cash Financing, the percentage of total offer price paid in cash, to capture deals subject to the Revlon standard, which represents another measure of 4 Cain and Davidoff (2012) also avoid using litigation data from SDC Platinum since its information is not reliable. 11

12 legal complexity (see Coates and Subramanian 2000) 5 ; (f) Target Termination Fee, an indicator variable equal to one for offers where a target pays a termination fee to a bidder if the deal is cancelled, as these deals are significantly associated with higher deal completion rates as well as higher takeover premia (see Bates and Lemmon 2003; Officer 2003); (g) Tender Offer, an indicator variable set equal to one for bidder tender offers for target shares, to capture extra Williams Act legal liabilities (Klein and Coffee 2000), different judicial standards, as well as target management incentives to be acquired (see Martin and McConnell 1991; Cotter, Shivdasani and Zenner 1997); (h) Bidder Minority Stake, an indicator variable for when the bidder holds a toehold position between 5% and 50% of target shares prior to the bid, and Controlling Shareholder Squeeze-out, an indicator for when the bidder holds a toehold position of 50% or higher, to capture unchallenged voting control and high expected control benefits (Officer 2003), stricter standards of judicial review and a higher likelihood of antagonistic entrenched target managers (see Betton, Eckbo and Thorburn 2009). C. Descriptive Statistics Table 1A examines all offers in our final sample, and shows that only 11.90% of all M&A offers are litigated % of all offers are completed 6 at an average takeover premium, measured relative to the target share price one week before the offer announcement, of about 37%. 7 The average offer size is $1.12 billion. 8 For litigated deals, 72.9% are completed at an average takeover premium of about 43% and their average offer size is about $3.9 billion (all of these numbers are significantly different from those of non-litigated offers). The yearly figures show that none of the offer features or outcomes are significantly different across the years. 5 Because of multicollinearity, we use either Stock Financing or Cash Financing variable in our regression analysis, but not both. 6 Bates and Lemmon (2003) report a similar completion rate of 79% for the period. 7 Moeller (2005) reports an average takeover premium, computed with respect to target s share price 6 days before offer announcement, of a little over 30%, for deals completed in the 1990s. We find that the takeover premium computed based on target share prices one week before the offer announcement is not significantly different from those computed 1 day or 1 month before announcement; so, we use the 1-week before announcement takeover premium henceforth. 8 Moeller, Schlingemann and Stulz (2005) report a similar average offer size of $1.99 billion for the period. 12

13 Of course, the probability of litigation depends on the type of offer. Table 1B shows that while the proportion of controlling shareholder squeeze-outs and hostile offers are relatively low (little more than 4% of all offers), these offers are litigated with significantly higher frequencies than the full sample. More than 45% of all controlling shareholder squeeze-outs and more than 28% of all hostile offers are litigated. We also segregate all offers into large and small offers, as defined as offers above and at or below the median offer size of $80 million respectively. We find that, in our final sample of 299 litigation offers, 64 pertain to small offers. Thus, unlike some recent papers (e.g., Cain and Davidoff 2012), it is important to consider smaller offers as well as large offers when analyzing M&A litigation and its effects on deals. Table 1C reports that in our final sample of 299 M&A litigation cases, the vast majority, over 86%, are target shareholder class-action lawsuits, over 10% are individual action lawsuits, and the remainder are derivative action lawsuits, bidder and target initiated lawsuits and others. The number of derivative action suits is low because, as Thompson and Thomas (2004) note, shareholders prefer to bring class-actions to challenge mergers and acquisitions transactions because they avoid additional procedural barriers that are raised in derivative suits. The number of target firm initiated suits in our sample is very low because the evolution of anti-takeover mechanism combinations like poison pills and staggered boards already generate long delays in hostile bidders ability to gain control of a target company, thereby decreasing the benefits targets can now derived from filing deal delaying litigation. There are only two antitrust lawsuits both of which also involve piggyback suits, where a government enforcement action is closely followed by a private action seeking to recover damages. This is unlike the federal securities class-action arena, where such piggyback suits occur more frequently (Cox and Thomas 2003). Finally, about 7.7% of all litigated M&A offers (23 offers) in our sample are subject to more than one type of lawsuit (e.g., both target firm and target shareholder initiated suits). The year-by-year analysis reveals that the proportion of individual 13

14 actions, and bidder and target lawsuits declines significantly from 1999 to 2000, while the proportions of other suits remain roughly the same. Examining lawsuit jurisdictions, Table 1D reports that about 49% of cases are filed in Delaware, roughly 44% are filed in other state courts outside of Delaware and the remaining 8% are federal suits. Some scholars claim that as of the late 1990 s Delaware began losing M&A litigation to other jurisdictions (Armour, Black and Cheffins 2012). Our data provide the first definite benchmark for what the jurisdictional breakdown looks like at the start of that period. Several scholars have found a dramatic increase multi-jurisdictional litigation during the post financial crisis period (Cain and Davidoff 2012; Daines and Koumrian 2012). We find very few of these cases in our sample period; there are only 10 multi-jurisdictional lawsuits out of 299 lawsuits in our sample, where M&A bids are litigated in both state and federal courts or in different state courts. This indicates that a major change in litigation strategy occurred over this relative short time period. Table 1E shows that 32.9% of the suits filed in our sample period are settled. Of these settlements, 42.27% resulted in increased deal consideration and an additional 18.56% generated other substantive relief. In comparison, the lawsuit settlements rate is much higher at 71.6% in the subsequent period for large offers (Cain and Davidoff 2012), and in the post-2008 global financial crisis time period, the consideration payment rate for large offers is only 5% (Daines and Koumrian 2012). Table 2 examines the associations between lawsuit types and lawsuit jurisdictions. Table 2A decomposes lawsuits filed in each type of jurisdiction into lawsuit types, while Table 2B offers the converse view, and reports where each type of lawsuit is filed. Table 2A shows that shareholder class action lawsuits are the dominant form of litigation in all courts, and are more prevalent in state courts than federal courts, reflecting the fact that they generally involve corporate issues arising under state law. On the other hand, the vast majority of individual 14

15 action, and bidder and target firm initiated lawsuits are filed in federal courts. These cases frequently raise disclosure claims under federal law required to be heard in federal court. Table 2B shows that no target firm initiated law suit was filed in Delaware. This may reflect the fact that targets generally try to file lawsuits in the jurisdiction where their headquarters or major operations are located (almost always outside of Delaware given its small size) in an effort to find a friendly judge who is more likely to rule in their favor. Bidder suits, on the other hand, are more evenly distributed across courts, as some bidders seek to bring federal disclosure claims in those courts, while the remaining bidders litigate state law claims in those forums. IV. LITIGATION FEATURES AND OUTCOMES A. Univariate Analysis Overall, litigated offers are significantly larger, and significantly more likely to be hostile bids, multiple-bidder offers, tender offers, bids involving bidder stake in target firms, bids using a higher proportion of cash financing, target termination fee provisions, as explained in Krishnan, Masulis, Thomas and Thompson (2012). Further, consistent with Krishnan, Masulis, Thomas and Thompson (2012), litigated offers have a significantly lower completion rate (72.9% versus 79.4%) and a significantly higher average takeover premium in completed deals as compared to non-litigated offers (43% versus 37%) (other results not reported to conserve space). Table 3 compares the frequencies and mean values of major M&A offer features and offer and lawsuit outcomes by lawsuit jurisdictions. The first two columns show that lawsuits filed in Federal courts are significantly more likely to be related to hostile offers and multiple-bidder offers, because Federal law is most likely to be associated with the disclosure and timing issues in such M&A offers [see Phillips and Krishnamurthy (2001)]. Columns 5 and 6 show that hostile bids are litigated least often in state courts other than Delaware. 15

16 No lawsuit pertaining to control shareholder squeeze-out litigation is filed in Federal courts, perhaps because the stringent legal precedent provided under the Weinberger standard in Delaware makes it a more attractive venue for plaintiffs. Indeed, comparing the figures reported in Table 3, columns 3 and 4, and columns 5 and 6, we see that Delaware attracts a significantly higher proportion of lawsuits related to control shareholder squeeze-outs as compared to cases filed in other courts. About one-third of the M&A lawsuits end in settlement and lawsuit settlement rates are not significantly different for Delaware and federal court cases, or between Delaware and other state cases. The consideration paid upon lawsuit settlements is significantly higher for Delaware as compared to other state courts because, as noted above, Delaware attracts a significantly higher proportion of lawsuits related to control shareholder squeeze-outs as compared to cases in federal courts and other state courts. The last two columns of Table 3 show that multi-jurisdictional lawsuits pertain to significantly larger offers and multi-bidder offers, as these are likely to be the economically better cases attracting litigation in a wider variety of forums. About two-thirds of multijurisdictional cases that settle involve payments of monetary awards, more than other lawsuits, although the differences are not statistically significant. This is consistent with better cases attracting more litigation in a wider variety of forums. Columns 3 through 6 of Table 3 show that a significantly lower proportion of offers with termination fee provisions are litigated in Delaware as compared to other courts. This could be because the incidence of termination fee provisions itself is lower in Delaware corporations, perhaps because of Delaware Court s unwillingness to provide a bright line test for the appropriate level of breakup fees, which could mean that they can be attacked as preclusive defenses under Unocal. Prior literature has shown that termination fee provisions have a significant positive association with deal completion because such fees can be viewed as legal devices that help reduce the tax on acquisition-related investments [see Officer (2003)]. 16

17 Examining lawsuit outcomes, a significantly higher proportion of deals litigated in state courts are completed, compared to cases brought in Federal court. This suggests that litigation pursued in state courts may be less of an obstacle to deal completion than litigation pursued in federal court. In particular, we find that litigated offers in Delaware court are completed significantly more frequently as compared to litigated offers in all other courts, while offers subject to multi-jurisdictional litigation are completed significantly less frequently than other litigated offers. However, takeover premia in completed deals are significantly lower for litigation pursued in state courts compared to federal court, suggesting less pressure on bidders to raise takeover premia in state court cases. Overall, the pattern shown in Table 3 is consistent with federal cases (that tend to pertain more often to hostile or multi-bidder offers, but not to controlling shareholder squeezeouts) raising more substantial obstacles to deal completions which can only be overcome by paying higher takeover premia. B. Multivariate Analysis In our multivariate analysis, in addition to controlling for offer and litigation features, we also include β Y, a vector of year fixed effects, β I, a vector of 10 bidder industry fixed effects, based on the 10 broad Fama-French industry sectors, and β T, an indicator that takes the value of 1 for cross border offers and 0 for domestic offers. Since the explanatory variables and residuals may exhibit serial correlation because the sample includes multiple firm and industry observations, we cluster standard errors by industry to produce more accurate confidence intervals that Petersen (2009) finds are well-specified in such situations. 1. Probability of Offer Completion We next examine the probability of deal completion using the following logit model: 17

18 (1) Y = β Y + β I + β T + β 1 Delaware Suit + β 2 Federal Suit + β 3 Multiple Jurisdiction Suits + β 4 Shareholder Class Action Suits + β 5 Individual Action Suits +β 6 Derivative Action Suits + β 7 Bidder Suits + β 8 Target Suits + β 9 Ln(Offer Size) + β 10 Intraindustry Offer + β 11 Hostile Offer +β 12 Multiple-Bidder Offer + β 13 Controlling Shareholder Squeeze-out +β 14 Bidder Minority Stake + β 15 Stock Financing +β 16 Target Termination Fee + β 17 Tender Offer +ε, where Y is the dependent variable of interest, taking the value of 1 for Completed Deals and 0 otherwise. We control for different types of lawsuits as well as offer characteristics as these can affect lawsuit outcomes Although we control for offer characteristics and fixed effects in the above regression specification, it is still possible that lawsuits can be associated with specific deal outcomes simply because litigation occurs more frequently in certain types of offers where the observed outcomes are more likely. To control for this form of selection bias, we employ an instrumental variable (IV) simultaneous equations regression model over our full sample, using limited information maximum likelihood (LIML) estimation (see Juergens and Lindsey 2009), where Shareholder Class Action Suits that accounts for 86.6% of all offers that entail lawsuits, is the endogenous covariate. To be a valid IV, it should have the properties that while it strongly predicts shareholder litigation, it is unrelated to the deal outcomes being examined. However, all the offer characteristics that we examine can be argued to be related to one or both of the deal outcome variables, invalidating their use as IVs under the exclusion requirement. So, we use industry-related litigation patterns from the recent past to obtain valid IVs (see, e.g., Krishnan and Masulis 2013; Krishnan et al. 2012). In particular, we use the top 10 bidder and target industries that have the highest proportion of M&A offers (subject to a minimum of 25 M&A offers) subject to litigation, as the IVs. Economically, the choice of our IVs is justified because they are industries that have attracted substantial M&A litigation in the recent past and are more likely to continue to attract litigation (and shareholder litigation) in future. However, there is no compelling reason to expect past industry associations with 18

19 litigation to be related to current M&A deal outcomes other than through the litigation itself, especially in the presence of controls for major offer characteristics. We define High Litigation Bidder (Target) Industry as an indicator variable that takes a value of 1 in the current year for the top 10 bidder (target) industries representing 2-digit SIC industries with the highest proportion of M&A offers subject to litigation over the past 3 years. For example, to calculate this indicator for year 1999, we use SDC data for the period, and to calculate this indicator for year 2000, we use our hand-collected litigation data for 1999 and the data from SDC. We then use High Litigation Bidder Industry and High Litigation Target Industry indicator variables as IVs for Shareholder Class Action Suits. The simple correlation between these two variables over our sample period is only 23%, which justifies the use of both as IVs. We examine the statistical validity of the instruments by performing overidentification tests (see, e.g., Krishnan et al. 2011). The F-statistic for the joint significance of the two IVs is 12.7 for Shareholder Class Action Suits, which is above the critical value of 10 recommended by Staiger and Stock (1997). Thus, the IVs strongly predict shareholder litigation. The Anderson-Rubin test statistic for over-identification yields insignificant p-values for M&A outcomes (namely Completed Deals or Takeover Premium), after controlling for other offer characteristics including lawsuits. So we fail to reject the joint null that the IVs are uncorrelated with the error term, which supports excluding them from the second-stage equation. We conclude that these two IVs satisfy the exclusion requirement of a valid instrument. The economic justification for our IVs is found in their descriptive statistics. The top 10 industries for which the High Litigation Bidder Industry takes a value of one in our sample period include chemicals, machinery, electronics, transportation equipment (which are all found in Eckbo (1992) to be industries in which mergers are most frequently challenged), communications, utilities, banks, insurance, and financial companies, which are all regulated industries (Agrawal and Knoeber 1996) where M&A regulatory or execution risk is high 19

20 because of more demanding regulatory requirements, and durable goods, where there are a large number of competitors. The top 10 industries for which High Litigation Target Industry takes a value of one are similar, except that transportation equipment is replaced by instruments, where firms have high levels of sensitive proprietary information, and financial companies is replaced by heath care where deals are generally fraught with technical, regulatory and commercial risks. The first column of Table 4 reports the regression coefficients, and associated z-statistics (in parenthesis) based on heteroskedasticity-consistent standard errors, clustered at the industry level, using the model in equation (1) after controlling for selection bias. Consistent with univariate results, lawsuits filed in the Delaware Chancery Court have a significantly higher probability of being completed, while lawsuits filed in multiple jurisdictions have a significantly higher probability of deal failure, consistent with the view that a greater number of litigation forums raises the legal hurdles to deal success. The signs of the significant control variables are as expected, namely large and intraindustry offers are associated with a higher probability of deal completion, while the deal complexity variables -- hostile bids and multiple-bidder offers -- are associated with a lower probability of deal completion. In line with the results reported in Officer (2003), offers that include target termination fee provisions, and employ tender offers are associated with a higher probability of deal success. Shareholder class action lawsuits (see Krishnan et al. 2012) and individual action lawsuits are associated with a higher probability of deal failure. 2. Takeover Premium The second and third columns of Table 4 report the regression coefficients, and associated t-statistics (in parenthesis) based on heteroskedasticity-consistent standard errors adjusted for industry clustering, using equation (1), where the dependent variable is either 20

21 Takeover Premium in All Offers or, alternatively, Takeover Premium in Completed Deals. Again Shareholder Class Action Suits is the endogenous covariate. In line with the univariate results, suits filed in federal courts entail significantly higher takeover premia. Shareholder class action suits and target lawsuits are associated with significantly higher takeover premia both in the full sample of offers and in the completed deals. Hostile offers (because of target resistance) and multiple-bidder offers (because of their competitive nature) are both associated with significantly higher takeover premia. Consistent with the evidence in Officer (2003), we find offers that employ target termination fees and tender offers are associated with significant higher takeover premia. A larger proportion of stock financing is also associated with higher takeover premia, potentially because of market timing allegations or weak stock performance can result in higher target purchase prices and greater risk of being expropriated as minority shareholders of the bidder. 3. Additional Checks To check whether the significant relationships between lawsuit jurisdictions and deal outcomes document above are driven by the use of the IV-LIML model and the choice of IVs, Table A.1 in the Appendix presents the results of a simple logit regression model for the probability of deal completion and a simple OLS regression model for takeover premium. The deal completion rate continues to have a significant positive relation (at the 5% level) to the choice of Delaware courts, while Takeover Premium in All Offers and Takeover Premium in Completed Deals have significant positive relations (at the 5% and 10% levels, respectively) to the choice of Federal courts, although the z-statistics and t-statistics (for Logit and OLS regressions, respectively) are lower than those using the IV-LIML model, showing that selection bias does affect unadjusted regression estimates. We also estimate two variants of our main regression models: one with only offer characteristics (that is, without lawsuit type indicator variables, Shareholder Class Action Suits, 21

22 Individual Action Suits, Derivative Action Suits, Bidder Suits, and Target Suits), and another with only these lawsuit type indicators and lawsuit jurisdiction indicators. In either variation, the deal completion rate continues to have a significant positive relation (at the 5% level) to the choice of filing in Delaware court, while Takeover Premium in All Offers and Takeover Premium in Completed Deals have significant positive relations (at the 5% and 10% levels, respectively) to the choice of filing in Federal courts (results untabulated to conserve space). While we are confident of finding virtually all M&A suits filed in Delaware, we cannot be sure that we have found all mergers and acquisitions cases filed in other state and Federal courts, only that we have found all material mergers and acquisitions cases (as interpreted by issuers and their lawyers) filed in those courts. Moreover, we note that we have only 25 Federal suits in our sample, as opposed to 147 Delaware suits and 131 non-delaware stats suits. Thus, in Table A.2 in the Appendix, we simulate results using weighted least squares methodology assuming that assume we have only 50% of all Federal suits. The relationship between lawsuit jurisdictions and deal outcomes becomes stronger: Delaware is now significantly and positively associated with deal completion success at 1% significance level, while Federal courts are significantly and positively associated with takeover premium in both all offers and completed deals, at the 5% significance level. 4. Litigation Settlements and Consideration Payments upon Settlements One key question we examine is what are the factors that make it more likely that litigation will produce favorable settlements and cash payments to plaintiffs on settlement? Critics of shareholder litigation argue that representative litigation rarely produces any direct benefits for shareholders (Romano 1991). Our data permit us to examine this issue for each type of jurisdiction, both in terms of the likelihood of settlements and the possible increase in deal consideration realized in lawsuit settlements. 22

23 In analyzing settlements, we describe the M&A shareholder litigation process. These cases are litigated on an expedited basis so that they can progress on the same timeline as the associated M&A transactions. As the deal parties work through the possible deal terms, attorneys for the target and bidder also negotiate with plaintiff shareholder attorneys in an effort to achieve a settlement, or to otherwise dispose of the litigation prior to deal closing. Table 5 reports the regression coefficients, and associated z-statistics (in parenthesis) based on heteroskedasticity-consistent standard errors with industry clustering, using equation (1), where Settled and Cash Consideration Paid on Settlement are now the dependent variables. The regressions are estimated over all litigated offers, and all settled litigated offers respectively. Again Shareholder Class Action Suits is the endogenous covariate. Since the target initiated lawsuit indicator is perfectly correlated with the failure to settle indicator, we dropped this control from the regression specification. Lawsuit jurisdiction does not significantly affect settlement rates or the consideration paid upon settlements. Nevertheless, it is interesting to note the effects of our control variables, which to the best of our knowledge, is not reported in the extant literature. Lawsuits challenging controlling shareholder squeeze-outs have a significant positive correlation with monetary settlements. This reflects the strength of the underlying legal claims in these cases. Shareholder class action lawsuits and individual action lawsuits are less likely to result in settlements. As expected, litigated hostile bids are significantly less likely to be settled, whereas same industry litigated bids are more likely to be settled. In the second column, we add three interaction terms. The first, Controlling Shareholder Squeeze-out Delaware Suits, enables us to see if a settlement is more likely in a Delaware case against a controlling shareholder. We find that this coefficient, though positive, is not significant at the 10% significance level. So it cannot conclude that Delaware courts are more demanding in these cases than courts in other jurisdictions. The coefficient of the second interaction term, Controlling Shareholder Squeeze-out Delaware Suits Tender Offer shows that controlling 23