The law and finance of antitakeover statutes.

• Author: Catan, Emiliano M.
• Position: III. Categorization Problems through Conclusion, with footnotes and tables, p. 655-680

3. Categorization Problems

   Part II above and the Appendix provide examples of three finance studies wherein the supposed link between ATSs and managerial ownership, leverage, and bond prices, respectively, can be explained by the failure to control for key variables or by mere misspecifications. There are, of course, many other studies of ATSs that we do not review in such a detailed way. Parts III through V discuss three problems that, to our knowledge, affect all studies of ATSs. These problems, together with the knowledge that ATSs do not materially increase a target's ability to defend itself (as explained in Part I), make us very doubtful that the results derived in these studies are causally attributable to ATSs.

   As mentioned, categorization problems affect even the best of the existing finance studies. These problems relate to errors regarding the year in which a state adopted a particular statute, to errors regarding the state in which a firm was incorporated, and to errors regarding which firms a state's ATS actually covered. These errors, together with a very conservative assessment of the effects of poison pills, tend to generate massive mismeasurement of whether and when firms became subject to takeover protection. (110)

   The errors we identify cannot be dismissed as noise that merely results in less accurate regression estimates. Rather, they have systematic (nonrandom) effects on the categorization of firms that render any result unreliable. This Part discusses some of these systematic effects. (111) Parts IV and V will address biases generated by the failure of studies to take account of managerial ownership and by the endogeneity of the state of incorporation, respectively.

   The various categorization errors identified have systematic effects because the firms affected by these errors are not randomly selected. Assigning the wrong year to when an ATS was adopted, omitting a statute entirely, or ignoring pill validation statutes or case law clearly establishing the validity of poison pills generally affects all firms incorporated in a specific state. Similarly, errors regarding the state of incorporation and which firms were covered by a state's ATS affect specific types of firms: firms that decided to reincorporate or firms that are headquartered outside their state of incorporation.

   But firms incorporated in a specific state, firms that reincorporated, and firms headquartered outside their state of incorporation are not a random selection of firms. For example, outside of Delaware and Nevada, most firms incorporated in a state have their principal place of business in that same state. (112) Firms incorporated in a certain state thus resemble each other in their geographic location and sometimes in other ways, such as the industries in which they operate.

   We provide a few specific examples that illustrate some of the systematic effects of the categorization errors:

   * Firms erroneously categorized as never being treated are systematically smaller: Under Bertrand and Mullainathan's categorization, 331 of the 2391 firms in our comparison sample were incorporated in states that did not adopt an ATS. (113) However, almost half of these firms (152) were in fact incorporated in states that had adopted either a business combination or a pill validation statute. These miscoded firms are substantially smaller (their median book value of assets as of 1985 was $18 million) than the firms in the comparison sample (median assets $55 million). (114)

   * Firms erroneously categorized as being treated between 1989 and 1990 are concentrated in Ohio: Under Bertrand and Mullainathan's categorization, 76 firms in the comparison sample became "treated" between 1989 and 1990, mostly because Ohio passed its business combination statute in 1990. (115) Since Ohio had adopted a pill validation statute in 1986, we regard these firms as having become subject to takeover protection four years earlier. (116) Of the 72 firms that are wrongly categorized, 68 firms (9496) were headquartered in Ohio. Overall, Ohio-headquartered firms constitute 4.896 of our comparison sample.

   * Firms erroneously categorized as being treated between 1985 and 1986 are systematically larger and have lower managerial ownership: Under Bertrand and Mullainathan's categorization, 184 firms (all incorporated in New York) in our comparison sample were treated between 1985 and 1986 (as New York adopted its business combination statute). (117) However, 52 of these firms were headquartered outside of New York and thus did not become subject to the statute. These 52 firms were larger (median assets $103 million) and had lower median board and managerial ownership (10.8%) than the firms incorporated and headquartered in New York (median assets $33 million; median D&O ownership 18.996). (118)

   Because economic shocks may have differential effects on, for example, large firms, firms located in a specific area, or firms with low board and managerial ownership, these systematic errors affect the regression estimates more severely than measurement errors that are simply random. For example, if Ohio Firms suffered an economic shock in around 1990, the impact of the shock would be reflected in the estimate for the ATS variable. (119)

   Moreover, the categorization errors result in firms being systematically regarded as becoming subject to takeover protection at a later date than was actually the case. Figure 1 below describes the fraction of firms in our comparison sample subject to takeover protection at the beginning of each year, under the methodology used by Bertrand and Mullainathan and under our methodology. The differences are stark. The fraction of firms subject to protection rises from 7.7% to 47% for 1986 and from 18.5% to 64% for 1988. Under our methodology, only 0.7% of firms became subject to takeover protection during 1990 or 1991 and only 7.5% of firms did not become subject to takeover protection at all during the standard study period (1976 to 1995). Under Bertrand

   and Mullainathan's methodology, the respective percentages are 5.2% and 13.9%. Importantly, under our categorization, the timeframe over which the bulk of firms became subject to takeover protection is much more compressed: within five years, over 90% of firms were covered by a statute. This compressed timeframe would generally increase the likelihood that regression estimates will be tainted by omitted-variable bias, as the estimates will be more prone to ascribing to takeover protection the effects of concurrent economic shocks that had differential effects on treated and control firms.

   [FIGURE 1 OMITTED]

4. Failure to Take into Account Managerial Share Ownership

   There are strong reasons to believe that Firms with different levels of board and managerial share ownership (inside ownership) will differ in their response to a change in state-supplied takeover protection. First, large inside ownership provides a defense against a hostile takeover. If management, say, owns thirty percent of the target stock, it becomes hard for a hostile raider to acquire a majority, especially since management can raise its stake further once a hostile bid is announced. Second, it provides significant incentives to increase the value of the equity held by the board and management. To the extent, for example, that a decrease in the takeover threat induces management to run the firm less efficiently (as argued by some commentators) (120) or that an increase in the takeover threat induces management to pursue short-termism at the expense of long-term value (as argued by others), (121) large inside ownership should produce significant counterincentives. Reasonable minds may differ as to when inside ownership is large enough to significantly reduce the threat of a hostile takeover or to overpower the incentives created by the takeover threat. We regard, respectively, a 30% and a 20% ownership stake as reasonable cutoffs. Once the board and upper management owns at least 30% of the company's stock, we would regard the protection against a hostile takeover afforded by that ownership as so significant that the additional protection offered by other antitakeover devices is not material. And once the board and upper management own at least 20% of the company's stock, we would regard the incentives provided by that ownership as so significant that the additional incentives generated by the presence or absence of other antitakeover devices are not material. (122)

   We were able to recover information on inside ownership for 1807 firms in our comparison sample. In 30% of these firms, directors and executive officers owned at least 30% of the firm's shares. In 44% of the firms for which we could recover ownership data, inside ownership exceeded 20% of the firm's shares. (123)

   The effect of director and officer ownership does not simply attenuate the impact of ATSs across the board. It also renders the firms incorporated in states that never adopted a business combination statute a very poor counterfactual for those incorporated in states that eventually adopted such a statute. Among the first group of firms, the median D&O ownership was 28.2%, and the fraction of firms in which directors and officers held more than 30% of the shares was 47.6%. By comparison, among firms incorporated in states that eventually adopted a business combination statute, the median D&O ownership was only 15.2%, and the fraction of firms whose directors and officers held more than 30% of the shares was 27.5%.124 Firms incorporated in states that never adopted a business combination statute are also dramatically smaller than their peers. In our comparison sample, the average (median) book value of assets reported for 1985 by firms that never became subject to a business combination statute was $190.6 million ($20.7 million),125 while the respective values for firms that eventually became subject to a business combination statute were $903...