DOES DELAWARE ENTRENCH MANAGEMENT?

JOHN M. OLIN CENTER FOR LAW & ECONOMICS WORKING PAPER NO. 08-024 DOES DELAWARE ENTRENCH MANAGEMENT? MURALI JAGANNATHAN AND A. C. PRITCHARD THIS PAPER CAN BE DOWNLOADED WITHOUT CHARGE AT: MICHIGAN JOHN M. OLIN WEBSITE HTTP://WWW.LAW.UMICH.EDU/CENTERSANDPROGRAMS/OLIN/PAPERS.HTM

Murali Jagannathan AA210, School of Management Binghamton, NY 13902 MuraliJ@binghamton.edu (607) 777-4639 A.C. Pritchard * 1039 Legal Research University of Michigan Law School Ann Arbor, Michigan 48109 acplaw@umich.edu (734) 647-4048 This Draft: December 2008 Abstract Critics have charged that state competition in corporate law, which Delaware clearly dominates, leads to a race to the bottom promoting management entrenchment at shareholders expense. We present evidence here inconsistent with this hypothesis. Measures of director quality and governance mechanisms are higher in Delaware. Delaware s directors hire higher quality CEOs and they are more likely to terminate CEOs. Tenures of Delaware directors and CEOs are both lower than their counterparts in other states. In addition, contrary to claims that anti-takeover laws promote management entrenchment, we find that states that provide the greatest anti-takeover protection Ohio, Pennsylvania, Massachusetts, and Maryland do not have significantly different turnover rates from California, the state that arguably offers the least anti-takeover protection. JEL Classification: G30, G34, K22 Keywords: Corporate governance, CEO turnover, Investor Protection. * Corresponding author. The authors thank Keith Bishop, John Coates, Martijn Cremers, Rob Daines, Michael Klausner, Srini Krishnamurthy, Y.C. Loon, Guhan Subramanian, and participants at the 2008 American Law & Economics Association and Conference on Empirical Legal Studies annual meetings, the 2008 New York Accounting and Finance Forum, and the Innovation, Business, and Law Colloquium at the University of Iowa for helpful comments and suggestions on an earlier draft of this article.

1. Introduction In this paper, we examine the relation between state corporate law and management entrenchment. State corporate law, by long-standing tradition in the United States, determines most questions of internal corporate governance the role of boards of directors, the allocation of authority between directors, managers and shareholders, etc. while federal law governs questions of disclosure to shareholders annual reports, proxy statements, and periodic filings. Despite substantial incursions by Congress, most recently in the Sarbanes-Oxley Act, this dividing line between state and federal law persists, so state law arguably has the greatest influence on corporate governance outcomes, such as management tenure. Companies have a good deal of discretion in choosing their state of incorporation. The allocation to the states of primary authority over corporate governance, when combined with the internal affairs doctrine, which holds that courts must apply the law of the state of incorporation to corporate law disputes, has created a regime of issuer choice in state corporate law. Corporations are free to choose the law of the state that best suits the needs of their directors, managers and shareholders, without regard to where the corporation principally does business. States can compete to attract firms by offering the most attractive menu of corporate law rules. This competition for corporate charters is not just about state pride: States that attract incorporations are rewarded with tangible benefits in the form of charter fees. Of equal importance, incorporations are also likely to produce work for the state s lawyers, who may be an influential lobbying force (Macey, 2005). Critics of issuer choice argue that states compete for corporate charters by pandering to corporate managers. These critics charge that states are caught in a race to the bottom, catering to management by providing rules which promote management entrenchment at the 1

expense of shareholders (Cary, 1974). States win in this competition, according to this view, by leaving shareholders vulnerable to overreaching by corporate managers. These critics point to state antitakeover laws as evidence for their position, and, as a policy matter, generally favor federal incorporation (see, e.g., Bebchuk and Ferrell, 2001). On the other side, advocates for state control over corporate governance respond that competition between states for corporate charters generates a race to the top. According to this camp, managers are constrained by competition in the capital markets to offer shareholders the corporate law rules that offer costeffective means of constraining the agency costs inherent in the separation of ownership and control (Winter, 1977). Whether the race is to the top or the bottom, Delaware has clearly prevailed in the competition for corporate charters. That state draws a clear majority of the nation s largest public companies to incorporate under its corporate code, despite its relatively small population and share of the national economy. More recent work (Subramanian, 2002) suggests that the competition for corporate charters is largely bilateral: states compete with Delaware in an effort to retain corporate charters. Does Delaware prevail in this competition by providing laws that entrench management? Boards of directors are an obvious place to look for factors influencing management entrenchment. Do boards of Delaware corporations differ in a systematic way from boards of companies incorporated in other states? After controlling for certain firm characteristics, we find that directors of firms that choose Delaware, on average, have shorter tenure and they serve on smaller boards. Thus, Delaware directors appear to be less entrenched and less likely to suffer from collective action problems. Indeed, we find evidence that Delaware firms are able to attract higher quality directors, an often stated motivation for firms to incorporate in Delaware. 2

Delaware directors sit on more public company boards and receive more compensation. We posit that the director liability protections offered by Delaware may make it attractive for high-quality directors to serve in Delaware firms. We also find that Delaware firms attract more institutional investors than other firms. Institutional investors may prefer the predictability offered by Delaware courts. If institutions are effective monitors of management, the evidence is inconsistent with Delaware firms avoiding external monitoring. Institutional investors are likely to be particularly concerned with the role that directors play in hiring top management and monitoring them. Accordingly, we analyze the quality of CEOs that the (higher-quality) Delaware directors hire and the termination decisions they make. Similar to Rajagopal, Shevlin, and Zamora (2006), we proxy for CEO talent based on how frequently a CEO is mentioned in lists by popular media, and find that Delaware CEOs are significantly more likely to appear on these lists. We also find that Delaware firms are more likely to hire an outside CEO. Thus, Delaware directors are able to hire higher quality CEOs, possibly from a larger pool of CEOs who are willing to serve under more predictable Delaware law. We also find Delaware CEOs, like Delaware directors, have shorter tenures than their counterparts elsewhere. These measures of CEO quality do not support the managerial entrenchment hypothesis. Does state of incorporation have any relation to CEO turnover? We find no evidence that that firms incorporated in Delaware are less likely to terminate (involuntarily) their CEO; indeed, we find involuntary turnover to be significantly more likely in most specifications. Our other findings, however, suggest that Delaware s higher turnover rate is unlikely to be related to its generally low level of anti-takeover protection. We find that the stringent anti-takeover states of 3

Ohio, Pennsylvania, Massachusetts and Maryland (post 1998) (OPMM) do not have less CEO turnover than do firms in other states. In addition, the supposedly shareholder-protective jurisdiction of California does not have significantly higher CEO turnover. These findings are inconsistent with the hypothesis that anti-takeover laws promote management entrenchment. We proceed as follows. Section 2 discusses prior literature and provides background on the differences between Delaware s corporate law and that of other states. Section 3 describes our sample. Section 4 presents our empirical results. We conclude with a discussion of our results in Section 5. 2. Background and prior research 2.1. Relevance of State of Incorporation for directors Delaware does not compete on price: Delaware incorporation fees are generally higher than those charged by other states, and incorporating in Delaware does not produce any particular tax advantages. So does Delaware corporate law differ from that of other states in a way that is likely to affect management tenure? The differences between the Delaware General Corporation Law and its main competitor, the Model Business Corporation Act (adopted in over forty states) are slight, so doctrinal analysis yields few obvious clues. One clue to understanding Delaware s lead in incorporations may come from comparing Delaware and California s corporate codes. A good portion of Delaware s lead comes at the expense of California (Subramanian, 2002). There are a number of differences between Delaware and California law that may push firms headquartered in California to incorporate in Delaware, including voting rules, which may be important to firms planning rapid growth, and protection against personal liability for directors, which may be important to firms seeking high- 4

demand directors and CEOs. On the personal liability point, Delaware arguably compares favorably not only to California, but also to virtually every other state. Some scholars have also pointed to anti-takeover provisions as an important driver of incorporation choice. On this point, Delaware has adopted an intermediate position. Delaware stands between California, which arguably offers the least anti-takeover protections, and the OPMM jurisdictions, which offer the most. 2.1.1. Voting Rules Romano (1985) finds that firms are likely to reincorporate in Delaware before committing to a program of mergers and acquisitions. Delaware, with its doctrine of independent legal significance, gives corporations flexibility in structuring transactions and will not import procedures applicable to one type of transaction into another type. 1 This doctrine takes on practical importance in allowing acquiring corporations to avoid shareholder votes and appraisal rights in most circumstances. Delaware corporations can set up a holding company structure without a shareholder vote (Del. G. Corp. L. 251(g)), which allows Delaware corporations to complete acquisitions through a triangular merger using a subsidiary without triggering a shareholder vote or appraisal. California law, by contrast, affords voting rights to acquiring company shareholders not only in mergers, but also in asset and stock purchases (Cal. Gen. Corp. L. 181(b) & (c), 1200(b) & (c), 1201(a)), if stock is used to complete the acquisition, and triangular transactions, if there is sufficient dilution of the parent company shareholders (Cal. Gen. Corp. L. 1200(e) & 1201(a)). 2 California is also more generous in affording appraisal rights to 1 Heilbrunn v. Sun Chemical Corporation, 150 A.2d 755 (Del. 1959) (rejecting de facto merger claim). 2 Ohio also affords acquiring company shareholders the right to vote (Ohio Rev. Code Ann. 1701.83). 5

acquiring company shareholders (Cal. Gen. Corp. L. 1300). Thus, growing firms intent on making acquisitions might opt for Delaware law to eliminate voting procedures with their attendant expense and delay. 3 Celikyurt et al. (2008) show that newly public firms make acquisitions at a very rapid pace, so Delaware s voting rules may be an important incentive for companies choosing their incorporation status at the IPO stage, particularly if they anticipate rapid growth after going public. If firms choose Delaware in order to facilitate acquisitions, this is not an anti-takeover motive. 2.1.2. Liability Protections In most respects Delaware s corporate law is not markedly different from the codes of other states, with the notable exception of California. If corporate law does not differ very much between the states on other margins, directors may take a special interest in provisions protecting them from personal liability. On the other hand, under the corporate law of virtually every state, the combination of the business judgment rule, stringent demand requirements, and broad statutory exculpation provisions means that directors face vanishingly small probabilities of being held personally liable for their acts as directors. (Black, Cheffins & Klausner, 2006). 3 Firms incorporated in California are also subject to cumulative voting (Cal. Gen. Corp. Law 708), which is intended to afford minority representation on corporate boards. Cumulative voting is unlikely to have much effect for firms with widely-dispersed shareholder bases, and California allows publicly-traded firms to opt out of the provision. (Cal. Corp. Law 301.5). The opt-out is limited, however, to firms that are listed on the NYSE, Amex, or NASDAQ Global. This means that many smaller public companies the status that most companies are likely to have at the time of their IPO, when the choice of incorporation becomes salient will be subject to cumulative voting, unless they opt to incorporate in Delaware or elsewhere. They are unlikely to switch back to California when they grow large enough to be listed. California firms that do opt for incorporation in another state may not have completely escaped California s regulatory grasp. California law purports to impose a number of its voting its requirements on foreign corporations if they have sufficient minimum contacts with California (Cal. Corp. Law 2115). Delaware courts, at least, refuse to be bound by California s attempts to regulate the internal affairs of Delaware corporations. VantagePoint Ventures Partners 1996 v. Examen, Inc., 871 A.2d 1108 (Del. 2005). California courts, of course, are likely to see it differently. 6

Does Delaware law offer provisions that might particularly appeal to directors? Delaware law does not differ significantly from the Model Business Corporation Act on the question of the standard of care or the protections of the business judgment rule. The M.B.C.A. also tracks Delaware law closely on the question of liability exculpation for breaches of the duty of care; if anything the M.B.C.A. may be slightly more generous to directors (Compare M.B.C.A. 2.02 with Del. Gen. Corp. L. 102(b)(7)). Kahan (2006) finds that states that have not adopted a liability limitation are significantly less likely to retain firms headquartered in their states. Once again, California is the leading example. California exculpates directors from liability for duty of care violations, but reckless acts are not covered (Cal. Corp. Code 204(a)(10)(A)(iv)). Given the ease with which recklessness can be pleaded, this is a substantial limit on the exculpatory force of California s provision. Further enhancing directors liability exposure, California also excludes indemnification for those acts (Cal. Corp. Code 204(a)(11)). Delaware law, by contrast, is particularly generous on indemnification. Delaware directors who prevail in a lawsuit against them have a statutory guarantee of indemnity from the corporation for the expense of their defense, (Del. Gen. Corp. L. 145(c)), which may be considerable. States following the M.B.C.A. also provide for guaranteed indemnification, but that provision requires complete exoneration (M.B.C.A. 8.52), whereas Delaware requires indemnification for partial success. 4 Delaware may also be more generous in allowing companies to provide permissive indemnification. (Compare Del. Gen. Corp. L. 145(f) with M.B.C.A. 8.59.) These differences are muted, however, by D&O insurance policies, which go beyond indemnification in the range of conduct that can be covered. Such policies are universal, but they are subject to coverage limits that may not fully protect directors. 4 See Merrit-Chapman & Scott Corp. v. Wolfson, 321 A.2d 138 (Del. Sup. Ct. 1974). 7

Beyond the differences in exculpation and indemnification, Delaware may promise directors more subtle advantages. Kahan (2006) finds that firms are more likely to incorporate in states with high quality judicial systems and flexible corporate law rules, two characteristics for which Delaware is well known (Romano, 1985). Commentators suggest Delaware s experienced and expert judges who sit on its Court of Chancery may play an important role in protecting shareholder interests (Fisch, 2001). That role is necessarily muted, however, by the very low probability that a director will be held personally liable. Notwithstanding the slim chance that a director will be found liable, the experience and expertise of Delaware judges may allow them to play an important shaming role, rebuking outside directors for inattention to their duties even while excusing them from liability (Rock, 1997). The Delaware Supreme Court s recent Disney decision is a prominent example of this style of decisionmaking. 5 Moreover, the impact of Delaware judges is likely to be amplified by the attention given to their decisions by the media and legal academics. Directors may be signaling their quality by pushing their firms to incorporate in Delaware, thereby announcing a willingness to have at least their reputation be held publicly accountable to shareholders (Iacobucci, 2006). They do so, however, cognizant of the fact that they will not personally bear the consequences of suit. The costs of suit will be covered by the company s D&O policy, and any settlement will be paid from that policy or by the company. The low probability of liability suggests that experienced and expert judges are not important because they are likely to intervene frequently to protect shareholder interests, thereby inducing Delaware board members to act as faithful monitors. Instead, an alternative causal story would suggest quality judges are important because they are likely to give directors comfort that they will not face liability because the judges produce litigation outcomes that are 8

predictable (Romano, 1985). Delaware law does not give directors comfort by shielding them from litigation; it simply guarantees that when litigation is brought, the directors will not be held personally liable (Macey, 2005). The predictability of Delaware law is further bolstered by the large stock of precedents to which its courts can look in deciding cases. Delaware s combination of expert judges and relatively comprehensive precedent provides a relatively predictable body of law (which would give lawyers confidence when providing advice to their clients). And should Delaware s judges slip and do the unpredictable, directors of Delaware firms can be confident that the Delaware legislature will step in to correct the problem. Some scholars suggest Delaware s competitive advantage is tied, in part, to its small population, which ensures that franchise tax revenues will be a significant portion of its overall budget (15% of revenues in 2007). 6 This is a powerful incentive for legislative attentiveness to the topic of corporate law. When the Delaware Supreme Court did the unthinkable in Smith v. Van Gorkom 7 holding the directors personally liable the Delaware legislature quickly restored equilibrium by allowing corporations to eliminate money damages for duty of care violations in their charters (Del. Gen. Corp. L. 102(b)(7)). The Delaware legislature s swift overturning of Smith v. Van Gorkom actually had the effect of accelerating Delaware reincorporations (Moodie, 2004). Delaware created a shock to the system, and then benefitted from the ensuing uncertainty in the directors and officers insurance market by fixing the problem more swiftly than its peer states. Even if the probability of liability is low, this concern is likely to be salient for outside directors, who have limited ability to control the firm s litigation exposure. Among outside directors, directors who serve multiple firms are most likely to be concerned about the potential 5 In re Walt Disney Co. Derivative Litigation, 906 A.2d 27 (2006). 6 State of Delaware, Comprehensive Annual Financial Report (2007), available at http://www.state.de.us/account/cafr.shtml. 7 485 A.2d 858 (Del. Sup. Ct. 1985). 9

for personal liability because each additional board membership increases the threat of liability. And directors, after all, make the decision where to incorporate. The lawyers who advise those directors are also likely to find liability concerns salient, and lawyers are the most common instigators of reincorporation decisions (Romano, 1985). Moodie (2004) documents that Delaware reincorporations surge after Delaware adopts liability protections for directors. Heron and Lewellen (1998) find positive abnormal stock returns for firms reincorporating for the purpose of obtaining liability protections for directors, suggesting that shareholders are cognizant of the role of such protections in attracting quality outside directors. If Delaware law has particular appeal to outside directors, this may manifest itself in the characteristics of the boards of Delaware firms. 2.1.3 Anti-takeover Provisions A number of scholars have pointed to anti-takeover protection as relevant to the question of whether Delaware s domination in incorporation reflects a race to the top or a race to the bottom. For a sample of exchange-traded U.S. corporations, Daines (2001) finds that firms incorporated in Delaware have significantly higher Tobin s Q in 12 out of 16 years between 1981 and 1996, and are significantly more likely to receive a takeover bid and to be acquired. Daines argues that the Delaware effect may be related to what he sees as Delaware s relatively mild anti-takeover statute, which may minimize management entrenchment. 8 He documents that takeover activity is not lower in Delaware. Boulton (2008) finds that IPO firms backed by venture capitalist are more likely to incorporate in states with fewer antitakeover protections, and that those firms are more likely to be taken over in the five years post-ipo. Subramanian (2002) 10

and Bebchuk and Cohen (2003), however, both find that firms are more likely to incorporate in the state where their headquarters are located if that state has adopted anti-takeover statutes. The findings from this body of work, taken as a whole, suggest that firms may sort themselves based on their willingness to be taken over: particularly for firms incorporated in states with stringent antitakeover protection, incorporating in Delaware may be akin to putting a For Sale sign on the door of the company s headquarters. These scholars may be looking under the wrong lamppost for the explanation of Delaware s advantage in the competition for charters. Differences in anti-takeover statutes may have little practical effect. Firms committed to protecting management can create their own defenses in most states by adopting a poison pill, which is clearly valid under Delaware law. And firms at the IPO stage can make this defense effectively invulnerable by adopting a staggered board. After controlling for other factors that might influence choice of incorporation, Kahan (2006) finds no evidence that firms are likely to incorporate in states with anti-takeover statutes. Moreover, Delaware s dominant market share came long before the advent of antitakeover provisions. A substantial portion of the preference for anti-takeover provisions found by Subramanian (2002) and Bebchuk and Hamdani (2002) may be explained by the exodus from California. California does not provide any explicit anti-takeover statutes, although its corporate code does makes it very difficult to cash out minority shareholders (Cal. Corp. Code 1101(e)), which may provide some secondary anti-takeover effect. Unlike most anti-takeover provisions, however, this provision is not subject to waiver by the target company s board. Avoiding this provision may push firms headquartered in California to incorporate in Delaware; this is not an 8 Daines result of a higher Tobin s Q for Delaware firms is called into question, however, by Subramanian (2004), who finds no Delaware effect for large firms between 1991 and 2002, and a disappearing Delaware effect for 11

anti-takeover motive. California also stands out in that the validity of the poison pill has not yet been established there; the pill may run afoul of that state s provision precluding discrimination among shareholders (Cal. Corp. Code 203). In Delaware, by contrast the validity of the pill is firmly established, although there are limits on the type of pill that can be adopted. Most states, however, provide more statutory anti-takeover protection than Delaware. 9 At the extreme are Ohio, Pennsylvania, Massachusetts, and (as of May 1999), Maryland. Ohio and Pennsylvania have statutes that force disgorgement of short-term gains by hostile bidders (Ohio Rev. Code Ann. 1707.043; 15 Pa. Cons. Stat. 2571-2575.); Massachusetts imposes classified boards on companies by statute (Mass. Gen. Laws Ann. Ch. 156B, 50A) and Maryland allows boards to adopt an effective classified board structure without shareholder authorization (Md. Code Ann., Corps. & Ass ns 3-803). A number of recent studies have attempted to measure the quality of corporate governance either by simply counting the number of anti-takeover statutes in a state, or by constructing corporate governance indices (Gompers et al. (2003), Bebchuk, Cohen, and Ferrell (2004)). As Bebchuk and Cohen (2003) note, however, counting the number of anti-takeover statutes may fail to provide an accurate picture of management protection because they do not account for the effect of substitute, non-statutory protections and they also may fail to weight smaller firms, i.e., greater value between 1991 and 1996, but no significant difference between 1997-2002. 9 If the quest for anti-takeover protection were the primary motivation for fleeing California, Delaware seems an unlikely destination: the neighboring state of Nevada not only has a statutory language validating poison pills (Nev. Rev. Stat. 78.195(5), 87.350(4), & 78.378(3)), but also gives director greater discretion in redeeming pills than Delaware does (Nev. Rev. Stat. 78.139). Moreover, Nevada not only has a business combination statute (Nev. Rev. Stat. 78.438) with fewer exceptions than Delaware s (Del. G. Corp. L. 203), but unlike Delaware, it has a control share statute (Nev. Rev. Stat. 78.379). In addition to Nevada s relatively stringent anti-takeover protections, it is also cheaper than Delaware, both in terms of franchise fees, and in terms of potential litigation costs. Taking all of these factors together, it seems unlikely that California firms choose Delaware incorporation for anti-takeover reasons. 12

appropriately substitute protection mechanisms. 10 For these reasons, Bhagat and Bolton (2006) argue that the measurement error in these indices can make them less informative than just using one or two variables that can be measured more accurately. They find that firms with higher governance indices are less likely to have forced management turnover. Based on this finding, they argue that these governance indices may be poor measures of entrenchment. 11 2.2. CEO Hiring and termination Decisions There is a substantial body of literature focusing on CEO turnover, but none of these papers focus on the role of state of incorporation. Falaye (2007) notes, in tabulated results, that the Delaware incorporation dummy is positively related to turnover, but does not discuss the issue further. We rely on this prior literature for the choice of control variables used in analysis below. A number of papers examine the impact of internal and external governance mechanisms on CEO turnover (e.g., Weisbach (1988), Huson, Parrino, and Starks (2001)). These papers consider these internal control mechanisms as exogenous. By contrast, we posit that firms may voluntarily choose to incorporate in Delaware because features of its corporate law may allow them to attract higher quality directors. If Delaware directors differ from the directors of 10 For example, the Gompers G-index counts, with equal weights, six anti-takeover statutes, despite the fact that such statutes vary widely in the protection that they offer from takeover. 11 Core, Guay and Rusticius (2006) find no causal relationship between these measures of weak governance and subsequent long-run performance, casting doubt either on the effectiveness of governance to affect returns or the effectiveness of these indices to measure governance. In addition, different corporate laws may be better or worse for different corporate constituencies. In particular, restrictions on payouts to shareholders may appeal to creditors, potentially decreasing a firm s cost of debt. Wald and Long (2007) find that firms incorporated in states with tighter payout constraints, including California (Cal. Corp. Code 500) and New York (N.Y. Bus. Corp. Law 510), carry lower levels of debt than do firms incorporated in Delaware, which does not impose a fixed payout constraint (Del. Gen. Corp. L. 170). Mansi et al. (2006) find that firms from states that restrict payouts have better credit ratings and lower yield spreads. They also find antitakeover statutes reduce the cost of debt for investment-grade firms. These results are consistent with the findings of Klock et al. (2004). Francis et al. (2006) find that Delaware firms pay higher bond spreads than firms incorporated in Ohio, Pennsylvania, or Massachusetts. Chava et al. (2008) find that Delaware firms pay higher interest for bank loans. Consistent with these findings, after controlling for certain governance 13

companies incorporated in other states, Delaware boards may be more willing to terminate underperforming CEOs. A number of papers consider the disciplinary role of takeovers; this work documents significant changes in management after such external control transactions (e.g. Martin and McConnell (1991), Denis and Denis (1995)). A standard agency cost theory would posit that discouraging takeovers may lead to lower sensitivity of turnover to performance. We are not aware of any papers that test the proposition that state-level takeover protection leads to a lower likelihood of turnover. 3. Data Description 3.1. Identifying historical State of Incorporation We obtain state of incorporation from Compact Disclosure and quarterly Compustat. We then check for reincorporation when we find a change of state of reincorporation and verify it with a reincorporation proposal in proxy statements. This process helps us identify our sample of incorporations and reincorporations. 12 For most of the study, we restrict our incorporation sample to Execucomp firms on Compustat with CRSP share codes of 10 or 11. We classify firms two ways: (1) by location of their headquarters (from Compustat); and (2) by state of incorporation. We sort the sample firms into four categories of states: (1) California; (2) Delaware; (3) Ohio, Pennsylvania, Massachusetts and (post-1998) Maryland, 13 for which we use the shorthand OPMM ; and (4) Other States. We classify California separately because of the large number of firms in our sample from that state, its low level of characteristics, Bradley et al. (2007), find that Delaware firms have lower credit ratings. Overall, these papers suggest that the choice to incorporate in Delaware may carry with it a higher cost of debt. 12 Mansi, Maxwell, and Wald (2006) collect this information from Mergent online. 14

anti-takeover protection, as well as the idiosyncratic features of its corporate law discussed in the last section. We classify separately firms incorporated in Ohio, Pennsylvania, Massachusetts, and Maryland based on the strength of those states anti-takeover statutes, which are generally regarded as a step above those found in other states, including Delaware. If anti-takeover statutes promote management entrenchment, the effect should be most pronounced for those four states, and least pronounced for California. Other States provide our baseline. We do not separately classify firms headquartered in Delaware because they are so few in number. 3.2. Identifying Governance Characteristics Data for most of the governance variables used in this study are obtained from the IRRC database. From here, we collect information on director tenure, number of directorships held by the directors, CEO-Chair separation, ownership by the CEO, percent of outside directors, and board size. Institutional ownership is also obtained from IRRC. Where that information is not available, we use the institutional ownership information from Compact Disclosure, if available. We obtain information on CEO tenure, Director Retainer, and Number of Meetings using Execucomp. 3.3. Identifying CEO Turnover We collect our primary data set relating to CEO turnover from Execucomp, which provides data on S&P 1,500 firms. Execucomp provides executive names and the date they become CEO which we use to identify CEOs and their turnover. We then use Factiva to search and classify these turnovers as Forced or otherwise. Based on the methodology in Parrino (1997), news reports which state the CEO was fired, forced out, or suggesting irreconcilable 13 Maryland s classified board statute was adopted in May 1999, so we include Maryland firms in this category only after the statute s adoption. 15

differences are classified as forced turnover. If the CEO leaves to join another firm, if they die, or if they retire, the turnovers are classified as unforced. For our robustness check, we collect CEO names and ages from Compact Disclosure for the period 1992-2004. From this list, we identify a forced CEO turnover as the year in which there is a change in the name of the CEO and the CEO is less than 60 years of age at that time. 16

3.4.Sources for other data Variable definitions are provided in the Appendix. We get all returns measures and the delisting codes from CRSP. All accounting data are from COMPUSTAT. We use Andrew Metrick s website for the G-index used as a measure of corporate governance. 14 We obtain the E-index of entrenchment from Lucian Bebchuk s website. 15 4. Empirical Results 4.1. Descriptive Statistics Table 1, Panel A summarizes trends in reincorporation for public firms from 1993 to 2004. Three features of the flow of corporations during that period stand out: (1) California is a substantial net loser of incorporations, with a net loss of 96 firms; (2) Delaware is a substantial net gainer, with a net gain of 207 firms; and (3) the OPMM jurisdictions are net losers, but the magnitude is much smaller than California s (net loss of 12 firms). <<Table 1 here>> These trends in reincorporation are confirmed in the makeup of our incorporation sample. Panel B of Table 1 breaks down our sample by state where the firm s headquarters is located and incorporation. Consistent with prior work, we find that Delaware is the overwhelming winner in the competition for charters in our sample, with Delaware incorporated firms representing 58% of the sample. California is again the conspicuous loser. Although firms headquartered in California make up 15% of our sample, firms incorporated in California constitute less than 14 http://www.som.yale.edu/faculty/am859/webpages/data.html. 15 http://www.law.harvard.edu/faculty/bebchuk/data.shtml. 17

3%. 16 For our sample, 78% of the firms headquartered in California opted for Delaware incorporation. By contrast, 44% of OPMM firms and 56% of firms headquartered in other states opted for Delaware incorporation. Table 2, Panel A provides descriptive statistics on firm characteristics that may influence choice of incorporating state and firm performance. For each of our three categories, we compare firms opting to incorporate in the same state as their headquarters with those that choose Delaware. The legal differences identified in the prior section suggest that firms incorporated in the state where their headquarters are located may differ systematically from those that opt to incorporate in Delaware. <<Table 2 here>> For our sample of Execucomp firms, we find that firms that choose Delaware over OPPM or Other states tend to have characteristics commonly associated with growth firms: lower book to market ratio, lower operating income with greater volatility of operating results, lower dividends and payouts, greater research and development expenditures, and fewer years have passed since their initial public offering. These differences are less pronounced when comparing California headquartered firms that opt for Delaware over California incorporation. Delaware firms do carry more debt than their California counterparts, which may reflect Delaware s laxer constraint on corporate payouts. Delaware firms tend to have greater total assets than firms incorporated in California, but are smaller than those incorporated in OPMM. This suggests that abandoning California for Delaware may be prohibitively costly for firms below a certain size (or not having the expectation of a substantial increase in size). Delaware firms are smaller at the median than their 16 Despite the relatively more stringent anti-takeover protections offered by Nevada, and the relative geographical proximity, only about 2.8% of California firms incorporate in Nevada, Thus, takeover protection is unlikely to be 18

Other state counterparts. The OPPM firms stand out as having the profile of more mature firms, with greater total assets, more share repurchases and higher dividend levels, lower research and development, and lower insider ownership. We next look at differences in governance characteristics. The discussion in the prior section suggests that incorporation in Delaware might allow a firm to attract a different kind of director. If Delaware firms attract higher quality directors, then this may translate into Delaware boards hiring higher quality CEOs and more effectively monitoring them. Panel B of Table 2 provides univariate statistics comparing the governance characteristics of firms incorporated in Delaware with those incorporated in their headquarters state. The boards of Delaware firms are generally smaller, but less independent. Boards of firms incorporated in the other jurisdictions may increase board independence by adding outsiders, rather than subtracting insiders from the board. We find no significant difference in the frequency of board meetings. Looking at the characteristics of the directors, directors of Delaware firms receive larger retainers than directors for California and Other state firms, perhaps suggesting higher demand for their services. The differences between OPMM and Delaware directors are not statistically significant. Greater compensation for Delaware directors does not reflect entrenchment: Delaware directors have a significantly shorter tenure than directors of firms incorporated elsewhere. We find additional evidence of higher demand for Delaware directors. We find that the outside directors of the Delaware firms in our sample serve on a significantly greater number of boards than do the directors for the California, OPMM, and all firms overall, and that this the primary factor determining incorporations. 19

translates into a greater likelihood that Delaware boards are Busy, defined as a majority of outside directors holding three or more directorships. Panel C of Table 2 presents evidence on shareholder influence. Delaware firms have significantly greater institutional ownership, suggesting they may face greater external scrutiny. Delaware firms have a significantly higher G-Index than California firms, but lower than OPMM and Other firms. This pattern holds as well for the E-Index, which focuses on structural antitakeover features. Both indices rely in part on the law of the state of incorporation, so these patterns are expected. To assess these differences in governance characteristics and shareholder influence more rigorously, we include Delaware incorporation as an independent variable in regressions using the governance characteristics as the dependent variable. Few models of governance structure exist in the literature motivating the relationship between firm characteristics and measures of governance. So, for all our models, we include the same control variables that are likely to be related to governance needs: Book to Market, Log Total Assets, and Years Since IPO. The results of these regressions, which account for clustering at the firm level, are presented in Table 3. The coefficient estimate for the Delaware variable is significant at the one percent level for a number of the governance characteristics when we use the entire sample (Panel A), including Board Size ( ), Director Retainer (+), Multiple Directors (+), Busyness (+). Most notably, Director Tenure is almost a year shorter for Delaware firms, which does not suggest entrenchment. Looking at shareholder influence, Institutional Ownership is nearly 5% higher in Delaware firms, suggesting that they realize the oft-stated motive of attracting institutional investors when they reincorporate to Delaware. G-Index ( ) and E-Index ( ) are also significant. When we restrict the sample to just firms with California headquarters (Panel 20

B), the Delaware variable is significant only for the regressions using Director Retainer and Multiple Directors as the dependent variables. Notably, the sign on the G-Index and E-Index flips, reflecting the low scores for California corporate law. The difference of about three units on the G-index and one unit on the E-index is very similar to the univariate differences, suggesting that differences in these indices may just be a function of state of incorporation rather than firm characteristics. The results for firms with OPPM headquarters (Panel C) are the same as the entire sample, with the exception of Director Retainer, which is no longer significant. <<Table 3 here>> Some of the director characteristics can be argued as promoting entrenchment or as measures of higher director quality, and it is difficult to differentiate the two arguments. For example, directors of Delaware firms tend to hold a larger number of directorships, which also results in boards that are more likely to be classified as busy. Ferris, Jagannathan, and Pritchard (2003) present evidence that firm performance affects the number of directorships held by an individual, which they call the reputation effect. The reputation of directors is closely tied to the performance of the firms that they serve, so the market for directors creates an incentive to promptly terminate a CEO in the face of poor performance. Critics (e.g., Fich and Shivdasani 2006), however, contend that directors who serve on multiple boards may have their attention spread too thin, making them ineffective monitors. Critics might also point to the higher compensation of Delaware outside directors as indicia of a lax monitoring environment. On the other hand, one would expect higher quality directors to be paid more. For the variables with relatively uncontroversial predicted directions, the results suggest that Delaware firms are less entrenched. Delaware boards are smaller, which research has shown makes for more effective monitoring (Yermack, 1996). Delaware outside directors have also 21

served for a shorter period of time, on average, which may mean that they are more independent. Turning to shareholder influence, Delaware firms have higher institutional ownership and lower G- and E- Indices for the overall sample. This finding suggests that attracting institutional investors may be a motive to incorporate in Delaware. This point is supported by anecdotal evidence from corporate lawyers, who say that they counsel clients to reincorporate in Delaware before their IPOs because Delaware law provides a known quantity for investors attempting to evaluate the firm. Incorporating in Delaware allows those investors to economize on information costs, which may be important if they have a larger number of portfolio companies. The heavier concentration of institutional investors is relevant to this study because institutional investors may exert pressure on board members to be active monitors of management performance. Standard agency cost theory posits that larger shareholders may be able to overcome collective action problems and keep management on a tighter leash. Del Guercio et al. (2008) find that shareholder activism is associated with greater CEO turnover. Overall, the evidence favors lower director entrenchment and more external monitoring in Delaware firms than firms incorporated in California, OPMM, or in other states. We look next at the other side of the balance, the CEO. Table 4 provides descriptive statistics comparing Delaware CEOs with CEOs from firms in other states. Panel A provides evidence on the supply of CEOs and their reputation. <<Table 4 here>> A reputation for quickly terminating CEOs in response to poor performance might dissuade some candidates from taking the CEO position (Murphy and Zabonjnik, 2007). We find, however, that Delaware firms are more likely to hire their CEOs from outside the firm, 22

although this difference is insignificant for the firms with California headquarters. 17 The protection against liability discussed above may be a factor allowing Delaware firms to attract outsiders as executives. If Delaware firms enjoy a higher supply of executives, this might lower their costs from terminating their CEOs for poor performance (Hermalin, 2005). We also find some evidence that Delaware CEOs may be higher profile, defined as being included in various Top N lists published by business periodicals. When compared with the entire sample, Delaware incorporated firms are more likely to have a high profile CEO; the differences are insignificant for California and OPMM. We find additional evidence that Delaware CEOs may have greater bargaining power. Panel B of Table 4 presents descriptive statistics on CEOs longevity and authority. For California headquartered firms, Delaware CEOs hold a greater number of outside directorships, perhaps suggesting more contacts. (The differences are insignificant for OPMM and Other states.) These greater contacts are not the result of longer service: like Delaware directors, Delaware CEOs have shorter tenure than their counterparts in other states (7.52 vs 8.26 years), although the differences are insignificant for California and OPMM. Delaware CEOs also tend to be younger. Despite their shorter service, Delaware CEOs have characteristics that may make it more difficult to dislodge them. CEOs of Delaware firms have greater share voting authority than their OPMM and Other counterparts, perhaps suggesting that they are more likely to be founders or members of the founding family. Moreover, Delaware firms are more likely to combine the position of CEO and Chair, which is likely to enhance the CEO s job security. This greater likelihood of combining the CEO and Chair positions in Delaware firms may reflect greater 17 We determine a CEO to have been hired from outside the firm if the Became CEO date in Execucomp is within one year of the Date Joined firm. 23

bargaining power wielded by outsider candidates to become CEO, or of a higher quality CEO candidate. Either way, they are less likely to be removed involuntarily. As with the differences in governance characteristics and shareholder influence, we include Delaware incorporation as an independent variable in regressions using the CEO characteristics as the dependent variable. We again use as control variables Book to Market, Log Total Assets, and Years Since IPO. The results are presented in Table 5. <<Table 5 here>> When we run the regressions for the entire sample, we find significant negative coefficients for the Delaware indicator variable for the regressions with CEO Tenure. The CEO Tenure coefficient is also negative for the California and OPMM sub-samples, but these coefficients are not significant. Delaware CEOs are also younger, except when compared to the OPMM firms, where the sign is reversed, although the coefficients are insignificant. The coefficients for CEO/Chair are insignificant. We also find a positive and marginally significant Delaware coefficient in the CEO Directorships regression for the California sub-sample. How do these differences in governance, shareholder influence, and CEO characteristics translate into the governance outcome of terminating the CEO? Table 6, Panel A summarizes the proportion of CEOs turned over in each of our three incorporation groups, presenting both forced turnover and the overall (forced + voluntary) turnover for our sample. The forced turnover rates in Delaware incorporated firms is about 2.6% and is significantly higher than the 1.93% rate observed for all non-delaware incorporated firms. Overall turnover is also higher for the Delaware firms, but the difference seems largely attributable to the higher forced turnover for Delaware firms. <<Table 6 here>> 24

To check whether some of the turnover rate differences documented in Panel A can be attributed to turnovers related to mergers or delistings, we compile turnover statistics for all firms that drop out of Execucomp. Panel B of Table 6 reports descriptive statistics for these companies. Number of firms refers to the number of unique firms on the Execucomp database during the 1993-2004 period. The last row provides the fraction of firms (not firm years) at which the CEOs lose their jobs due to mergers or delistings. In a large fraction of mergers, the target CEO remains part of the combined firm for about a year. For the turnover calculation in Panel B, these Not Forced turnovers are included. The pattern of turnover observed with mergers and delistings is consistent with that of turnovers identified from surviving firms in our sample: CEOs of Delaware firms are more likely to be turned over than their counterparts in firms incorporated in other states (24% vs. 19%). 18 4.2. Regression Analysis The univariate statistics of Table 6 suggest that Delaware firms are more likely to terminate their CEOs. The analysis above, however, suggests that Delaware firms differ from their counterparts in their financial, governance, shareholder, and CEO characteristics. Accordingly, in this section we use a multivariate framework to examine the relation between Delaware incorporation and the likelihood of forced turnover. We use two approaches to this question. In the first, we use Forced Turnover as the dependent variable, which equals one if the CEO is turned over in that year and zero otherwise. 19 Panel A of Table 7 presents the results from single stage logistic regressions. Panel B presents 18 The multivariate analysis presented in Table 7 contains only of turnovers described in Panel A of Table 6. It does not include the turnover related to mergers and delistings. 25

the results from a two stage approach similar to Jenter and Kanaan (2008). In the first stage we decompose the returns in a year into systematic component based on industry returns and firmspecific component. In the second stage, we estimate the CEO turnover regression using the Cox proportional hazard model. In this model, we estimate the probability that the CEO will be fired in a given year, conditional on (s)he continues to be employed after last year. In these regressions, we consider voluntary turnovers to be right-censored. 20 <<Table 7 here>> In the logistic regressions presented in Panel A, we include two variables for firm performance. The two firm performance variables are stock market returns for one and two years prior, adjusted by industry (using Fama/French 48 industry classification). We industryadjust our performance measures to filter out exogenous industry shocks. 21 The standard errors presented in the table account for potential clustering at the firm level. In the base regressions, which includes only the two performance variables and indicator variables for Delaware, California, and OPMM incorporation, the coefficient for the Delaware indicator variable is 0.266 (Odds ratio 1.31), which is significant at the 5% level. The coefficient for California incorporation is negative, but insignificant, which does not support the hypothesis that California s supposedly pro-shareholder corporate law discourages management entrenchment. Notably, the coefficient for the OPMM indicator variable is also insignificant, which does not support the hypothesis that the stringent antitakeover protections in those states entrench management. It is possible, however, that takeovers and terminations are substitute disciplinary 19 We also instrumented Delaware incorporation using Book to Market, Operating Income Volatility, R&D/Sales, and Operating Income. The results obtained from the instrumental variable probit regressions are qualitatively the same as in the single stage regression. 20 Given the small proportion of forced CEO turnovers each year, we use annual observations instead of monthly observations as in Jenter and Kanaan (2008). If we use monthly observations, the Delaware variable is stronger than the results discussed here. 21 The results are qualitatively unaltered if we use raw returns. 26

mechanisms for dealing with poorly performing CEOs. The termination rate may be boosted in the OPMM firms because of the lower likelihood of takeover. The other models presented in Panel A include characteristics of the board, shareholder influence, and the CEO that may affect the likelihood of termination. We saw in the previous sections that Delaware firms differ from their counterparts from other jurisdictions with respect to a number of these characteristics. In all but the model with board control variables, the coefficient of Delaware remains significantly positive and is similar in magnitude to the coefficient in the base regression. A number of characteristics relating to the board are significant, but only Board Meetings (+) and Director Tenure ( ), are significant in the full model. The positive coefficient for Board Meetings suggests that the directors of troubled firms have been monitoring the firm more closely prior to the turnover. The variables relating to shareholder influence (Institutional Ownership and E-Index) are also insignificant in the full model, although E-Index is significant in the shareholder characteristics model. 22 The coefficient, however, is positive, the opposite of the predicted direction. Among the CEO characteristics, CEOs with longer tenure and who serve as Chair are less likely to be terminated for poor performance relative to industry peers. Overall, Delaware firms appear to be more likely to terminate CEOs, which is inconsistent with the argument that Delaware laws entrenches managers. For our second set of regressions, we run a hazard model similar to the turnover regression in Jenter and Kaanan (2008). In the first stage, we run regressions of firm s returns in the turnover year (Year 1) and in the previous year (Year 2) on the corresponding period equally weighted industry (Fama-French 48 industry classification) returns. These regressions predict the probability that the current CEO will be fired over the next year. The results, 27

presented in Panel B of Table 7, are quite similar to those from the logit models. In the base models, the Delaware coefficient is significant at the five percent level. When we include the board control variables, the coefficient for the Delaware incorporation variable is no longer significant at conventional levels, and the pattern of significance is essentially the same for the control variables, except that High CEO Voting is significant with a negative coefficient in the CEO model. In three of the five models the Delaware indicator variable is significant at the five percent level; the two models that include board characteristics generate insignificant Delaware coefficients. Overall, we find no support for the managerial entrenchment hypothesis posited by the race to the bottom proponents. Our findings suggest that Delaware firms are more likely to terminate CEOs for poor performance than firms incorporated in other states. Moreover, we find no evidence that firms incorporated in shareholder protective California are more likely to terminate their CEO for poor performance. Nor do we find evidence that the antitakeover OPMM states are any less likely to terminate CEOs for poor performance. 4.3 Robustness Checks Tables 1-7 present results for the Execucomp sample of S&P 1500 firms. As a test of robustness for our central findings, however, we do a similar analysis for a larger sample of Compact Disclosure firms. For this larger sample, we define Forced Turnover as CEO departure before age 60, a relatively noisy measure which may limit the power of our tests relative to the more refined data that we have for the Execuomp sample. We also get a limited number of governance measures from Compact Disclosure: Board Size, Percent Outside Directors, Insider Ownership, Institutional Ownership and CEO Age. Despite the noise in these measures, the 22 We get similar results when we substitute the broader G-Index for the E-Index. 28

Delaware indicator variable is positive and significant, consistent with the results presented earlier. 23 4.4 Implications Our central findings are that the directors of poorly performing Delaware firms are significantly more likely to terminate the CEO. A greater likelihood of terminating an underperforming CEO, although inconsistent with management entrenchment, does not necessarily translate to effective board monitoring. Directors of Delaware firms may be overreacting to performance that is beyond the control of the CEO (Fisman et al., 2005). Equivalently, firms incorporated in less stringent anti-takeover states may perform better due to threat of external monitoring, and any observed poor performance may be more likely due to bad luck rather than the poor quality of the manager. It is also possible that the truly effective directors fire the CEOs before the onset of any poor performance. Do Delaware boards overreact in firing their CEOs too often? Terminating a CEO is costly and disruptive; it is unlikely to yield better performance if the board is unable to locate a better successor. To explore the possibility that Delaware boards may be overreacting, we analyze whether the long-term operating performance and stock performance of Delaware firms is poorer than firms in other states following management turnover. In unreported regressions, we analyze changes in operating performance after turnover of the CEO for the Delaware, OPMM, California, and Other firms. Operating performance significantly improves for all groups after forced CEO turnover, but the differences between the groups are insignificant. The results are similar when we analyze the long run stock performance after forced CEO turnover. 23 Altering the age threshold to 61 or 62, or redefining turnovers to include managers whose stock was delisted due to performance reasons or whose firms were acquired, does not alter the principal results. 29

The alphas from Fama-French three factor and Carhart four factor regressions are positive for all groups. But a hedge portfolio formed by shorting Delaware firms and holding all other firms long from the month of the turnover and holding it for one year (or two), does not produce significant alphas. This indicates that turnover in Delaware is not due to Delaware directors willing to pull the trigger on CEOs too quickly. 5. Conclusion This study focuses on the relation between a firm s state of incorporation and its corporate governance. We find that Delaware firms attract directors who enjoy greater demand for their services and more institutional investors. Our primary measure of the efficacy of corporate governance is the likelihood that a firm terminates its CEO after poor performance. In most of our specifications, we find that Delaware firms are more likely to fire their CEO in response to poor firm performance. Nonetheless, this great willingness to terminate underperforming CEOs does not appear to limit the ability of Delaware firms to attract external CEO candidates. Nor do we find evidence that competition among states in the provision of corporate law promotes director or CEO entrenchment; Delaware directors and CEOs have a shorter tenure than their counterparts in firms incorporated in other states and are subject to more institutional monitoring. Overall, we conclude that our findings are inconsistent with the race to the bottom hypothesis. We find no evidence that competition among the states in the provision of corporate law promotes management entrenchment or director tenure. Delaware may be pandering to management, but it does not appear to do so by entrenching management. 30

References Bebchuk, L.A., A.Ferrell, 2001. A new approach to takeover law and regulatory competition. Virginia Law Review, 87, 111-164. Bebchuk, L.A., A.Hamdani, 2002. Vigorous Race or Leisurely Walk: Reconsidering the Competition over Corporate Charters. Yale Law Journal, 112, 553-615. Bebchuk, L.A., A. Cohen, 2003. Firms Decisions Where To Incorporate. Journal of Law and Economics, 46, 383-425. Bebchuk, L. A., A. Cohen, Allen Ferrell, 2004. What Matters in Corporate Governance? Review of Financial Studies, forthcoming. Bhagat, S., B.Bolton, 2006. Corporate Governance Indices. Working Paper, University of Colorado. Black, B., B.Cheffins, M.Klausner, 2006. Outside Director Liability. Stanford Law Review 58, 1055-1159. Boulton, T. 2008.Venture Capital and the Incorporation Decisions of IPO Firms, Working Paper, Miami University. Bradley, M., D.Chen, G. Dallas, E. Snyderwine, 2007. The Relation between Corporate Governance and Credit Risk, Bond Yields and Firm Valuation, Working Paper, Duke University. Cary, W., 1974. Federalism corporate law: reflections upon Delaware. Yale Law Journal, 83, 663-705. Celikyurt, U., M. Sevilir, A. Shivdasani, 2008. Going Public to Acquire? The Acquisition Motive in IPOs. Working Paper, University of North Carolina. Chava, S., D. Livdan, A. Purnanandam, 2008. Do Shareholder Rights Affect the Cost of Bank Loans? Working Paper, Texas A& M. Coates, J., 2000. Takeover Defenses in the Shadow of the Pill: A Critique of the Scientific Evidence on Takeover Defenses. Texas Law Review, 79, 271-382. Daines, R., 2001. Does Delaware law improve firm value? Journal of Financial Economics, 62, 525-558. Del Guercio, D, L. Seery, T. Woidtke, 2008. Do boards pay attention when institutional investor activists just vote no? Journal of Financial Economics, 90, 84-103. Denis, D.J., D.K. Denis, 1995, Performance Changes Following Top Management Dismissals, Journal of Finance, 50, 1029-1057. Faleye, O.2007. Classified boards, firm value, and managerial entrenchment, Journal of Financial Economics, 83, 501-529. Ferris, S.P., M.Jagannathan, A.C. Pritchard, 2003. Too Busy to Mind the Business? Monitoring by Directors with Multiple Board Appointments, Journal of Finance, 58, 1087-1111. Fich, E.M., A. Shivdasani, 2006. Are Busy Boards Effective Monitors? Journal of Finance, 61, 689-724. Fisch, J.E., 2001. The Peculiar Role of the Delaware Courts in the Competition for Corporate Charters. University of Cincinnati Law Review, 68, 1061-1100. Fisman, R., R.Khurana, M.Rhodes-Kropf, 2005. Governance and CEO Turnover: Do something or do the right thing? Working Paper, Columbia University. Francis, B.B., I.Hasan, K.John, M. Waisman, 2006. Does Delaware Incorporation Affect the Firm's Cost of Debt Capital? Working Paper, Rensselaer Polytechnic Institute. 31

Gompers, P., J.Ishii, A.Metrick, 2003. Corporate Governance and Equity Prices, Quarterly Journal of Economics, 107-155. Hermalin, B., 2005. Trends in Corporate Governance, Journal of Finance, 60-5, 2351-2384. Heron, Randall A., Wilbur G. Lewellen, 1998. An Empirical Analyssi of the Reincorporation Decision, Journal of Financial and Quantitative Analysis, 23, 549-568. Huson, M.R., R.Parrino, L.T. Starks, 2001. Internal Control Mechanisms and CEO Turnover: A Long-Term Perspective, Journal of Finance, 56, 2265-2297. Iacobucci, E.M., 2006. Toward a Signaling Explanation of the Private Choice of Corporate Law. American Law and Economics Review, 6, 319-344. Jenter, D., F. Kanaan, 2008, CEO turnover and Relative Performance Evaluation, Working Paper, MIT. Kahan, M., 2006. The Demand for Corporate Law: Statutory Flexibility, Judicial Quality, or Takeover Protection? Journal of Law, Economics, and Organization, 22, 340-365. Klock, M.S., S.A. Mansi,.W.F. Maxwell, 2004. Corporate Governance and the Agency Cost of Debt, Working Paper, University of Arizona. Macey, J., 2005. Delaware: Home of the World s Most Expensive Raincoat, Hofstra Law Review, 33, 1131-1139. Mansi, S.A., W.F. Maxwell, J.K. Wald, 2006. Do State Laws Matter for Bondholders? Working Paper, Virginia Tech. Martin, K.J., and J.J. McConnell, 1991. Corporate Performance, Corporate Takeovers, and Management Turnover, Journal of Finance, 46, 671-687. Moodie, G., 2004. Forty Years of Charter Competition: A Race To Protect Directors from Liability? Working Paper, Harvard University. Parrino, R,, 1997. CEO turnover and outside succession A cross-sectional analysis, Journal of Financial Economics, 46, 165-197. Rajgopal, S., T. Shevlin and V. Zamora, 2006. CEOs Outside Employment Opportunities and the Lack of Relative Performance Evaluation in Compensation Contracts, Journal of Finance, 61, 1813-1844. Rock, E.B., 1997. Saints Sinners: How Does Delaware Corporate Law Work? UCLA Law Review, 44, 1009-1108. Romano, R., 1985. Law as a Product: Some Pieces of the Incorporation Puzzle. Journal of Law, Economics, and Organization, 1, 225-283. Subramanian, G., 2002. The Influence of Antitakeover Statutes on Incorporation Choice: Evidence on the Race Debate and Antitakeover Overreaching, University of Pennsylvania Law Review, 150, 1795-1873. Subramanian, G, 2004. The Disappearing Delaware Effect. Journal of Law, Economics, and Organization, 20, 32-59 Wald, J. K., M. S. Long, 2007. The Effect of State Laws on Capital Structure, Journal of Financial Economics, 83, 297-319. Weisbach, M., 1988. Outside directors and CEO turnover. Journal of Financial Economics, 20, 431-460. Winter, R.K., 1977. State law, shareholder protection, and the theory of the corporation. Journal of Legal Studies, 6, 251-292. Yermack, D., 1996. Higher market valuation of companies with a small board of directors, Journal of Financial Economics, 40-2, 185-211. 32

APPENDIX: VARIABLE DEFINITIONS Variable Forced Turnover All Turnover CA HQ CA Inc. DE Inc. OPMM HQ OPMM Inc. All firms HQ = Inc All firms Inc = DE Total Assets Book to Market Leverage Dividends Total Payout R&D Operating Income Description Indicator variable equal to one if newspaper reports indicate that CEO was forced out, if the board mentions dissatisfaction with firm performance, if there are disagreements between the CEO and the board of directors, or if the departure is unexpected and no particular reason was provided for the departure; zero otherwise. Source is Factiva news searches. Indicator variable equal to one if the CEO is different from the prior year; zero otherwise. Source is Execucomp. Firm with headquarters in California. Firm incorporated in California. Firm incorporated in Delaware. Firm with headquarters in Ohio, Pennsylvania, Massachusetts, and Maryland (post 1998) headquarters. Firm incorporated in Ohio, Pennsylvania, Massachusetts, and Maryland (post 1998). All firms on sample incorporated in the state of their headquarters All firms on sample that are incorporated in Delaware In millions. Ratio of the book value of equity to the market value of equity (shares outstanding x price at end of the year) Ratio of long term debt plus the debt in current liabilities scaled by total assets. Dividends paid to common shareholders, scaled by earnings available to common shareholders. Sum of dividends and stock repurchases, scaled by earnings available to common shareholders. Amount of research and development, scaled by Sales. Operating income, scaled by total assets. 33

Income Volatility Capital Expenditures Years Since IPO Yr. 1 Ind. Adj. Ret. Yr. 2 Ind. Adj. Ret. Board Size Percent Outsiders Institutional Ownership G-Index E-Index Board Meetings Director Retainer Director Tenure Multiple Directors Busyness CEO Voting High CEO Voting Power CEO Directorships Standard deviation of operating income over prior five years. Investment in long term assets, scaled by total assets. Number of years since firm s initial public offering. Operating income for year prior, scaled to mean return for industry. Operating income for year two years prior, scaled to mean return for industry. Total number of directors. Percentage of outside directors on the board. Percentage of shares held by institutions. Source is IRRC; if not available on IRRC, the data are obtained from Compact Dislosur/Spectrum. Governance index from Gompers, Ishii, & Metrick (2003). When the G-index is not available for a particular year, we use the nearest past years index as our measure of the current year s G-index. Entrenchment index from Bebchuk, Cohen, & Ferrell (2004). Average number of meetings per year held by the full board for the prior three years. Cash compensation paid to directors. Mean number of years that outside directors have been on board. Mean number of external directorships of public companies held by outside directors. Indicator variable equal to one if a majority of outside directors hold three or more directorships, and zero otherwise. Percentage of votes held by CEO. CEO holding more than five percent of voting power. Number of outside directorships held by CEO. 34

CEO-Chair CEO Age External CEO High Profile CEO Indicator variable equal to one if the CEO also serves as Chair, and zero otherwise. Age of CEO in years Indicator variable equal to one if CEO named to that position within one year of joining the firm, and zero otherwise. Indicator variable equal to one if CEO is included in one of the following lists, and zero otherwise (Rajagopal, Shevlin, and Zamora, 2006). Source List Years Business Week The Best Managers 1992-2004 Financial Times World s Most Respected Business Leaders 1998,1999,2002 Financial Times Most respected companies 1998-2003 Time Magazine Time/CNN 25 Most Influential 2001 Fortune 50 Most Powerful Women in Business 1998-2004 Fortune Most Powerful Black Executives in America 2002 All variables are computed at the beginning of the year. Financial variables are from Compustat and governance variables are from Execucomp and IRRC, except for G-Index, which is obtained from Andrew Metrick s website, http://www.som.yale.edu/faculty/am859/webpages/data.html, and the E-Index, which is obtained from Lucian Bebchuk s website, http://www.law.harvard.edu/faculty/bebchuk/data.shtml. Ownership data is from Compact Disclosure. 35

Panel A: Reincorporations from 1993-2004 Table 1 Reincorporations and Incorporation The Table provides the frequency of firms initial state of incorporation and the state to which they reincorporate. Data is collected from Factiva news searches for the years 1993-2004 and verified in proxy statements. Reincorporations from and to a foreign country, and reincorporations prior to the IPO are excluded. Reincorporation CA DE OPMM OTH Total Incorporation CA 0 96 0 7 104 DE 6 0 21 82 117 OPMM 1 31 2 7 43 OTH 1 197 8 34 244 Total 8 324 31 130 508 Panel B: Headquarters and Incorporation The Table shows the frequency of firms headquarters state and the state in which they incorporate. The sample includes all firm-years for which CEO data is available on Execucomp from1993-2004. Only firm years with positive total assets on Compustat and CRSP share code of 10 or 11 are included in the sample. Foreign firms incorporated in the US are excluded. Incorporation Headquarters OPMM OPMM Other Other CA DE (own) (other) (own) (other) Total CA 513 2,494 25 158 3,190 DE 0 83 6 6 95 OPMM 0 1,346 1,557 49 134 3,086 Other 41 8,049 219 4,991 1,115 14,415 Total 554 11,972 1,557 299 4,991 1,413 20,786 36

Table 2 State of Incorporation and Financial and Governance Characteristics The Table provides the means and medians (in parentheses) of various firm and governance characteristics classified by the firm s incorporation choice. The sample consists of all Execucomp firm years with positive total assets on Compustat during the period 1993-2004 that have a CRSP share code of 10 or 11. The tests of differences column gives the t-values from a two sample t-test and z-values from Wilcoxon rank sum tests comparing the previous two columns. Panel A presents firm financial characteristics. All accounting ratios are winsorized at the 1% and 99% levels. Panel B presents governance characteristics. Statistical significance at the five percent levels and one percent levels are in italics and bold, respectively. Panel A: Firm Characteristics and Incorporation Choice CA HQ/ CA Inc. CA HQ/ DE Inc. Tests of Differences OPMM HQ/ OPMM Inc. OPMM HQ/ DE Inc. Tests of Differences All firms HQ = Inc. All firms Inc. = DE Tests of Differences Total Assets 3,219.60 (312.49) Book to Market 0.414 (0.339) Leverage 11.92 (5.28) Dividends 13.79 (0) Total Payout 31.23 (0) R&D 14.91 (5.45) Operating 18.26 Income (17.73) Income 8.41 Volatility (5.71) Capital 6.84 Expenditures (5.46) Years Since 15.50 IPO (9.56) 4,421.13 (529.90) 0.437 (0.351) 15.93 (9.60) 7.82 (0) 33.53 (0) 24.88 (7.20) 13.64 (14.91) 12.83 (6.82) 6.12 (4.53) 13.93 (10.01) 2.27 ( 4.28) 1.24 (0.09) 5.40 ( 4.03) 4.20 (5.35) 0.50 (0.48) 2.76 (-3.80) 4.38 (3.85) 5.55 ( 3.71) 2.48 (3.15) 2.04 (0.71) 5,332.24 (1004.19) 0.498 (0.430) 20.91 (19.11) 25.94 (17.59) 52.47 (34.67) 5.66 (0.06) 15.13 (13.93) 5.01 (3.02) 5.65 (4.70) 26.824 (24.58) 4,341.77 (750.52) 0.455 (0.386) 21.37 (19.58) 14.27 (0) 31.13 (7.83) 14.89 (0.90) 13.85 (15.11) 7.31 (4.21) 5.56 (4.66) 18.16 (11.20) 1.97 (4.25) 2.29 4.24 0.65 (0.81) 7.99 (11.33) 5.17 (7.99) 5.25 (4.28) 2.00 (0.73) 7.18 7.81 0.55 (0.80) 12.63 (15.36) 8,091.17 (1051.10) 0.527 (0.461) 22.73 (21.61) 27.17 (15.62) 50.26 (31.32) 5.22 (0) 15.99 (14.40) 5.14 (3.02) 6.51 (5.07) 24.69 (22.06) 8,689.22 (915.36) 0.478 (0.402) 22.91 (20.71) 15.60 (0) 39.91 (13.52) 9.58 (0) 15.04 (15.10) 7.80 (4.13) 6.60 (4.87) 18.01 (11.55) 1.10 (5.23) 6.86 (11.44) 0.67 (3.07) 21.84 (29.39) 6.50 (16.10) 6.13 ( 14.48) 3.57 ( 0.09) 10.37 ( 21.09) 1.04 (3.20) 26.12 (31.97) 37

Panel B: Governance Characteristics and Incorporation Choice CA HQ/ CA Inc. CA HQ/ DE Inc. Tests of Differences OPMM HQ/ OPMM Inc. OPMM HQ/ DE Inc. Tests of Differences All firms HQ = Inc. All firms Inc. = DE Tests of Differences Board Size 8.40 (7) 8.14 (8) 1.26 (0.03) 10.350 (10) 9.35 (9) 6.90 (5.99) 10.11 (10.00) 9.27 (9) 15.31 (15.29) Percent Outsiders 64.04 (66.67) 63.97 (66.67) 0.05 (0.47) 68.14 (71.43) 64.68 (66.67) 4.25 (4.71) 65.24 (66.67) 63.65 (66.67) 4.90 (5.23) Board Meetings 7.75 (7) 7.47 (7) 1.63 (1.26) 7.50 (7) 7.32 (7) 1.49 (2.19) 7.26 (7) 7.26 (7) 0.07 (0.40) Director Retainer 13.95 (13) 17.99 (18) 7.02 ( 5.47) 18.76 (17.5) 17.90 (18) 1.80 (1.09) 18.78 (18) 20.72 (20) 5.40 ( 10.07) Director Tenure 7.83 (7.10) 6.65 (6.29) 3.76 (3.12) 8.07 (7.75) 6.84 (6.13) 6.60 (7.91) 8.09 (7.60) 6.88 (6.40) 15.87 (16.55) Multiple Directors 0.71 (0.63) 0.94 (0.86) 4.44 ( 4.76) 0.96 (0.857) 1.03 (1) 1.99 ( 1.02) 0.91 (0.80) 1.01 (0.88) 6.32 ( 6.27) Busyness 13.13% 22.44% 8.78 18.68% 22.90% 3.79 18.15 24.05 48.83 Panel C: Shareholder Influence and Incorporation Choice CA HQ/ CA Inc. Institutional 52.6 Ownership (53.68) G Index 5.86 (5) E Index 1.08 (1) CA HQ/ DE Inc. 58.19 (61.13) 8.58 (8) 2.06 (2) Tests of Differences 3.93 ( 4.33) 20.42 ( 17.11) 15.88 (14.21) OPMM HQ/ OPMM Inc. 56.57 (58.73) 10.66 (11) 3.00 (3) OPMM HQ/ DE Inc. 61.15 (62.76) 9.17 (9) 2.38 (2) Tests of Differences 4.63 ( 5.14) 13.43 (12.82) 12.47 (12.29) All firms HQ = Inc. 54.55 (56.37) 9.63 (10.00) 2.61 (3) All firms Inc. = DE 59.23 (62.20) 8.99 (9) 2.24 (2) Tests of Differences 12.23 ( 13.79) 15.21 (16.02) 19.16 (19.45) 38

Table 3 Multivariate Comparison of Governance Characteristics, Shareholder Influence, and Incorporation The Table shows the results of regressions (clustered robust standard errors in parentheses) of certain governance characteristics as the dependent variable. The variables are defined in the Appendix. All regressions are estimated with OLS, except for Busyness, which is a estimated with a logistic regression. Panel A provides the results for regression using the entire sample, while Panel B is the subsample of firms with California headquarters that are incorporated in California or Delaware, and Panel C is the subsample of firms with OPMM headquarters that are incorporated in either their headquarters state or Delaware. Coefficients that are significant at the five percent level and one percent level are in italics and bold, respectively. Panel A: Entire Sample Board Size Intercept 2.206 (0.235) Book to 0.183 Market (0.070) Log Total 1.006 Assets (0.038) Years Since 0.014 IPO (0.003) DE Inc. 0.704 (0.100) Percent Outsiders 0.505 (0.016) 0.001 (0.006) 0.013 (0.002) 0.002 (0.000) 0.003 (0.007) Board Meetings 4.366 (0.195) 0.318 (0.078) 0.358 (0.029) 0.006 (0.003) 0.109 (0.093) Director Retainer 8.121 (0.995) 1.540 (2.073) 3.304 (0.155) 0.096 (0.016) 3.136 (0.559) Director Tenure 7.417 (0.366) 0.086 (0.121) 0.072 (0.048) 0.042 (0.004) 0.904 (0.161) Multiple Directors 0.256 (0.069) 0.121 (0.025) 0.143 (0.010) 0.005 (0.001) 0.137 (0.031) Busyness 4.259 (0.241) 0.601 (0.142) 0.344 (0.029) 0.011 (0.003) 0.462 (0.097) Institutional Ownership 42.616 (1.671) 3.439 (0.663) 1.908 (0.246) 0.009 (0.023) 4.885 (0.777) G-Index 7.556 (0.281) 0.100 (0.081) 0.165 (0.040) 0.029 (0.004) 0.431 (0.120) E-Index 2.345 (0.139) 0.157 (0.043) 0.014 (0.020) 0.003 (0.002) 0.363 (0.059) N 11,996 11,964 19,457 20,097 9,610 9,614 9,614 14,346 16,131 16,160 R 2 / Pseudo R 2 0.388 0.007 0.054 0.062 0.0658 0.1587 0.0877 0.040 0.080 0.028 39

Panel B: Firms with California Headquarters Board Size Intercept 1.730 (0.523) Book to Market 0.052 (0.311) Log Total 0.859 Assets (0.086) Years Since 0.038 IPO (0.009) DE Inc. 0.303 (0.317) Percent Outsiders 0.557 (0.041) 0.006 (0.021) 0.010 (0.005) 0.001 (0.001) 0.001 (0.024) Board Meetings 5.367 (0.541) 0.115 (0.223) 0.375 (0.085) 0.001 (0.012) 0.442 (0.345) Director Retainer 9.570 (2.166) 1.010 (0.763) 3.477 (0.334) 0.104 (0.043) 3.215 (1.150) Director Tenure 6.680 (0.994) 0.312 (0.493) 0.030 (0.123) 0.065 (0.016) 1.007 (0.654) Multiple Directors 0.087 (0.156) 0.191 (0.066) 0.099 (0.021) 0.000 (0.003) 0.225 (0.087) Busyness 3.117 (0.635) 0.361 (0.285) 0.186 (0.080) 0.002 (0.009) 0.656 (0.336) Institutional Ownership 34.498 (4.574) 8.283 (2.252) 3.731 (0.687) 0.101 (0.073) 4.927 (2.637) G-Index 3.756 (0.546) 0.105 (0.251) 0.185 (0.083) 0.045 (0.011) 2.853 (0.294) E-Index 0.867 (0.284) 0.364 (0.142) 0.016 (0.043) 0.010 (0.006) 0.969 (0.156) N 1595 1.593 2,765 2,869 1,311 1311 1,311 2,115 2,108 2,107 Adj-R 2 0.427 0.019 0.044 0.301 0.091 0.082 0.025 0.083 0.240 0.104 Panel C: Firms with OPMM Headquarters Board Size Intercept 1.143 (0.719) Book to Market 0.395 (0.142) Log Total 1.216 Assets (0.124) Years Since 0.010 IPO (0.009) DE Inc. 0.664 (0.266) Percent Outsiders 0.504 (0.043) 0.013 (0.015) 0.016 (0.006) 0.002 (0.000) 0.018 (0.018) Board Meetings 4.467 (0.518) 0.299 (0.196) 0.382 (0.088) 0.006 (0.007) 0.011 (0.244) Director Retainer 2.377 (2.145) 0.598 (0.861) 2.437 (0.380) 0.149 (0.034) 1.122 (0.997) Director Tenure 8.579 (0.943) 0.091 (0.304) 0.123 (0.128) 0.016 (0.010) 1.130 (0.392) Multiple Directors 0.012 (0.171) 0.012 (0.052) 0.093 (0.026) 0.009 (0.002) 0.180 (0.075) Busyness 4.332 (0.660) 0.059 (0.198) 0.283 (0.086) 0.019 (0.007) 0.526 (0.258) Institutional Ownership 53.079 (4.323) 5.115 (1.463) 0.701 (0.633) 0.042 (0.053) 4.837 (1.977) G-Index 8.356 (0.785) 0.249 (0.193) 0.168 (0.113) 0.031 (0.009) 1.174 (0.315) E-Index 2.679 (0.347) 0.184 (0.095) 0.012 (0.051) 0.005 (0.004) 0.604 (0.148) N 1,733 1,731 2,748 2,828 1,404 1,405 1,406 2,038 2,335 2,333 R 2 /Pseudo R 2 0.429 0.087 0.052 0.219 0.038 0.135 103.36 0.028 0.142 0.075 40

Table 4 State of Incorporation and CEO Characteristics Variable definitions are in the Appendix. Panel A provides the fraction of CEOs with certain characteristics. Panel B provides descriptive statistics on CEO characteristics that might influence the turnover decision. Tests of differences test for the difference in proportion or tests for the differences in means (medians), of firms incorporated in Delaware. The column either shows chi-square and p-values or the t-values (z-values) from a two sample t-test of means (wilcoxon rank-sum test). Panel A: CEO Supply and Incorporation Choice CA HQ/ CA HQ / Tests of OPMM HQ/ OPMM HQ/ Tests of All firms All firms Tests of CA Inc. DE INC differences OPMM Inc. DE Inc. differences HQ = Inc. Inc. = DE differences External CEO 59.87 60.29 0.02 42.52 52.69 13.59 40.27 46.74 41.00 High Profile CEO 3.42 3.25 0.04 0.66 0.73 0.05 1.85 2.37 7.28 Panel B: CEO Characteristics and Incorporation Choice CA HQ/ CA Inc. CA HQ / DE INC Tests of differences OPMM HQ/ OPMM Inc. OPMM HQ/ DE Inc. Tests of differences All firms HQ = Inc. All firms Inc. = DE Tests of differences CEO Directorships 0.33 (0) 0.52 (0) -3.64 (-2.47) 0.85 (0) 0.76 (0) 1.66 (1.65) 0.65 (0) 0.62 (0) 1.70 (1.63) CEO Tenure 8.58 (5.59) 7.82 (5.67) 1.80 (0.62) 7.96 (5.50) 7.93 (5.65) 0.10 0.45 8.26 (5.92) 7.52 (5.42) 6.92 (6.81) CEO Voting 3.80 (1.37) 3.57 (1.60) 0.46 ( 0.63) 3.21 (1.10) 6.29 (1.30) 4.66 ( 2.86) 3.93 (1.00) 4.50 (1.30) 2.91 ( 4.68) CEO-Chair 48.24% 58.93% 10.03 66.12% 70.95% 4.55 67.95 68.56 0.51 CEO Age 54.48 (54) 53.18 (53) 3.28 (3.28) 54.90 (55) 53.89 (54) 3.72 (4.13) 55.14 (55) 54.41 (55) 7.12 (7.02) 41

Table 5 Multivariate Comparison of CEO Characteristics and Incorporation This table shows the results (robust standard errors in parentheses) of regressions with certain CEO characteristics as the dependent variable. The variables are defined in the appendix. All regressions are OLS, except for CEO-Chair, which is logit. DE indicates incorporation in Delaware. Panel A provides the results for regression using the entire sample, while Panel B is the subsample of firms with California headquarters that are incorporated in California or Delaware, and Panel C is the subsample of firms with OPMM headquarters that are incorporated in either their headquarters state or Delaware. Coefficients that are significant at the five percent level and one percent level are in italics and bold, respectively. Panel A: Entire Sample CEO Directorships Intercept 0.413 (0.085) Book to Market 0.097 (0.024) Log Total Assets 0.122 (0.013) Years Since IPO 0.007 (0.001) DE Inc. 0.016 (0.037) CEO Tenure 9.112 (0.589) 0.231 (0.181) 0.021 (0.080) 0.023 (0.006) 0.865 (0.280) CEO Voting Power 11.728 (0.921) 0.146 (0.369) 0.934 (0.120) 0.032 (0.012) 0.341 (0.449) CEO Age 49.602 (0.520) 0.488 (0.185) 0.501 (0.071) 0.066 (0.007) 0.252 (0.256) CEO-Chair 0.679 (0.180) 0.118 (0.063) 0.155 (0.025) 0.012 (0.002) 0.115 (0.079) N 9,433 18,831 10,096 20479 11,926 R 2 / Pseudo R 2 0.090 0.006 0.036 0.057 0.026 42

Panel B: Firms with California Headquarters CEO Directorships Intercept 0.285 (0.168) Book to Market 0.148 (0.105) Log Total Assets 0.093 (0.025) Years Since IPO 0.001 (0.004) DE Inc. 0.191 (0.097) CEO Tenure 6.991 (1.759) 0.245 (0.547) 0.193 (0.245) 0.022 (0.027) 0.731 (1.056) CEO Voting Power 5.990 (1.757) 0.511 (0.884) -0.135 (0.309) 0.057 (0.032) 0.385 (0.957) CEO Age 50.449 (1.420) 1.584 (0.560) 0.282 (0.195) 0.132 (0.030) 1.665 (0.986) CEO-Chair 1.268 (0.463) 0.303 (0.218) 0.199 (0.068) 0.002 (0.008) 0.405 (0.287) N 1,286 2,787 1,373 2,948 1,592 R 2 / Pseudo R 2 0.042 0.007 0.016 0.085 0.022 Panel C: Firm with OPPM Headquarters CEO Directorships Intercept 0.133 (0.256) Book to Market 0.098 (0.057) Log Total Assets 0.100 (0.041) Years Since IPO 0.009 (0.003) DE Inc. 0.012 (0.106) CEO Tenure 9.583 (1.584) 0.729 (0.459) 0.030 (0.208) 0.053 (0.014) 0.536 (0.812) CEO Voting Power 11.859 (3.099) 0.487 (0.596) 1.013 (0.371) 0.044 (0.018) 2.471 (1.622) CEO Age 48.282 (1.245) 0.054 (0.426) 0.604 (0.179) 0.077 (0.014) 0.229 (0.603) CEO-Chair 0.535 (0.512) 0.051 (0.152) 0.087 (0.076) 0.020 (0.006) 0.414 (0.213) N 1,372 2,633 1,456 2,876 1,717 R 2 / Pseudo R 2 0.071 0.020 0.047 0.080 0.031 43

Table 6 CEO Turnover and State of Incorporation Panel A in table below shows the percentage of turnovers each year under each incorporation category. The sample consists of all Execucomp firms on Compustat and CRSP during the period 1993-2004. Turnovers are classified as forced using a methodology similar to Parrino (1997). Panel B provides the reasons for firms exiting the Execucomp database. Merger refers to the number of firms that are acquired where there are no reports of conflicts between the management of the acquirer and the target. When such disagreements are found, they are classified as hostile. Delisting refers to the number of firms that stop trading due to performance reasons. Turnovers in these firms are considered to be forced if the CEO is removed immediately at or before the merger/delisting announcement and there are news reports indicating poor performance. If the CEO is part of the transition team after the merger and serves for at most a year with the merged firm, then the turnover is considered to be Not Forced. If the target CEO continues as an officer of the acquirer for more than a year, then we classify it as No turnover. Panel A: Turnover CA HQ/ CA Inc. CA HQ/ DE Inc. OPMM HQ/ OPMM Inc. OPMM HQ/ DE Inc. All non-de Inc. All DE Inc. Forced Turnover All Turnover Forced Turnover All Turnover Forced Turnover All Turnover Forced Turnover All Turnover Forced Turnover All Turnover Forced Turnover All Turnover 1993 2.70 5.41 1.38 3.45 0.00 4.31 0.00 2.22 1.95 3.90 0.74 3.57 1994 0.00 2.63 2.55 7.01 0.83 3.31 4.30 7.53 1.69 7.02 1.61 6.66 1995 0.00 5.13 3.51 8.77 3.17 10.32 0.00 5.10 1.90 6.67 2.76 8.61 1996 0.00 2.00 3.65 8.33 4.13 5.79 4.72 5.66 1.44 4.71 3.40 7.72 1997 5.36 10.71 3.02 5.53 1.61 10.48 4.42 8.85 2.03 9.13 2.70 6.51 1998 6.25 10.42 2.35 8.92 3.20 9.60 4.27 5.13 2.33 8.16 2.22 7.23 1999 0.00 4.65 2.84 6.16 0.75 5.22 7.44 13.22 1.65 7.27 3.77 8.53 2000 0.00 5.41 3.11 8.00 2.26 11.28 5.31 11.50 2.93 9.82 3.62 10.15 2001 0.00 2.86 4.07 6.79 2.22 9.63 1.79 9.82 1.61 7.88 2.56 8.18 2002 2.78 2.78 1.81 7.24 2.94 7.35 3.57 9.82 2.45 8.06 2.53 7.29 2003 5.00 10.00 2.23 4.46 0.75 5.22 2.56 7.69 1.97 7.75 2.50 6.49 2004 0.00 5.13 2.27 4.09 1.57 6.30 0.88 4.39 1.29 7.75 2.36 7.17 All 2.01 5.82 2.54 6.59 1.83 7.44 3.29 7.73 1.93 7.35 2.60 7.40 Panel B: Mergers, delistings and turnover CA HQ/ CA Inc. CA HQ/ DE Inc. OPMM HQ/ OPMM Inc. OPMM HQ/ DE Inc. All non-de Inc. All DE Inc. Number of firms 62 365 190 184 1063 1642 Merger 17 96 47 41 267 446 Hostile 0 2 0 1 2 4 Delisting 0 8 1 5 7 46 No turnover 3 17 12 7 72 97 Not Forced 13 82 36 37 193 371 Forced 1 7 0 3 11 28 Turnover / # firms (%) 22.58 24.38 18.95 21.74 19.19 24.30 44

Table 7 Determinants of Forced CEO turnover The table presents models of the likelihood of forced CEO turnover, using Forced Turnover as the dependent variable for each. Panel A presents the results of single stage logistic regressions estimated with clustered (robust) standard errors. In panel B, we run a hazard model similar to the turnover regression in Jenter and Kaanan (2008). In the first stage, we run regressions of firm s returns in the turnover year (Year 1) and in the previous year (Year 2) on the corresponding period CRSP value weighted returns. For the hazard model, voluntary turnovers are considered as censored. Coefficients for DE Inc. that are significant at the five percent level and one percent level are in italics and bold, respectively. To maximize data usage in the full models presented on the right hand side of each table, we set a value of zero to variables with missing values, and correspondingly set a value of 1 to an indicator variable signifying that the variable data was missing. Coefficients for these indicator variables are not reported in the table; they are generally significant at the five percent level or lower. Panel A: Single Stage Base Shareholder Board CEO Full Coeff. SE Coeff. SE Coeff. SE Coeff. SE Coeff. SE Intercept 4.125 0.096 4.125 0.316 4.187 0.369 3.320 0.411 4.108 0.581 Yr 1 Ind. Adj. Ret. 1.079 0.105 1.090 0.109 0.892 0.102 1.071 0.111 0.942 0.109 Yr 2 Ind. Adj. Ret. 0.888 0.095 0.897 0.100 0.796 0.098 0.869 0.099 0.819 0.105 DE Inc. 0.266 0.114 0.303 0.103 0.174 0.108 0.266 0.100 0.220 0.109 CA Inc. 0.028 0.340 OPMM Inc. 0.028 0.188 Log Total Assets 0.062 0.033 0.090 0.038 0.059 0.033 0.100 0.041 Board Size 0.027 0.026 0.039 0.028 Percent Outsiders 0.009 0.351 0.142 0.372 Board Meetings 0.159 0.011 0.153 0.012 Director Retainer 0.001 0.000 0.001 0.001 Director Tenure 0.068 0.021 0.066 0.020 Multiple Directors 0.196 0.091 0.172 0.096 Institutional Ownership 0.002 0.003 0.002 0.003 E-index 0.118 0.041 0.091 0.047 CEO Directorships 0.096 0.078 0.112 0.083 CEO Te nure 0.037 0.010-0.023 0.010 High CEO Voting Power 0.514 0.272 0.324 0.294 CEO-Chair 0.517 0.124 0.652 0.142 CEO Age 0.001 0.007 0.002 0.008 Pseudo R 2 0.0543 0.0607 0.1113 0.0741 0.1348 Wald Chi-Sq. 199.95 229.35 481.83 287.19 580.56 N 19,522 19,522 18,885 19,450 18,818 45

Panel B: Hazard Model Base Shareholder Board CEO Full Coeff. SE Coeff. p-value Coeff. p-value Coeff. p-value Coeff. p-value Predicted Year 1 Return 0.998 0.138 1.060 0.147 0.826 0.145 1.014 0.139 0.897 0.147 Predicted Year 2 Return 0.844 0.135 0.812 0.140 0.667 0.139 0.874 0.137 0.673 0.142 Residual Year 1 1.557 0.126 1.545 0.133 1.198 0.130 1.545 0.130 1.246 0.133 Residual Year 2 1.235 0.124 1.247 0.131 1.052 0.127 1.227 0.126 1.055 0.131 DE Inc. 0.236 0.110 0.242 0.101 0.159 0.104 0.229 0.098 0.154 0.105 CA Inc. 0.022 0.188 OPMM Inc. 0.057 0.316 Log of Total Assets 0.057 0.030 0.061 0.034 0.032 0.030 0.072 0.036 Board Size 0.030 0.025 0.037 0.026 Percent Outsiders 0.064 0.348-0.054 0.363 Board Meetings 0.114 0.008 0.105 0.008 Director Retainer 0.002 0.001 0.001 0.001 Director Tenure -0.061 0.021 0.058 0.021 Multiple Directors 0.120 0.093 Institutional Ownership 0.004 0.003 0.003 0.003 E-index 0.062 0.046 0.004 0.021 CEO Directorships 0.060 0.068 0.089 0.072 High CEO Voting Power 0.655 0.250 0.479 0.267 CEO-C hair 0.523 0.120 0.591 0.129 CEO Age 0.005 0.006 0.004 0.007 N 19,522 18,131 18,885 19,501 18,865 Wald Chi-Sq. 329.87 318.57 655.87 391.76 721.20 46