Which Antitakover Provisions Matter?

Jonathan M. Karpoff is Washington Mutual Endowed Chair in Innovation and Professor of Finance at the University of Washington Foster School of Business; Robert J. Schonlau is Assistant Professor of Finance at Brigham Young University Marriott School of Business; and Eric W. Wehrly is Assistant Professor of Finance at Western Washington University. This post is based on their recent paper.

Related research from the Program on Corporate Governance includes What Matters in Corporate Governance? by Lucian Bebchuk, Alma Cohen, and Allen Ferrell.

Researchers disagree sharply over which antitakeover provisions affect a firm’s takeover likelihood. Some argue, for example, that poison pills are the only important takeover deterrents, while others claim that classified boards or coverage by business combination laws are. Other researchers argue that golden parachutes deter takeovers, while some contend that they facilitate takeovers. Similarly, there is significant disagreement about whether such provisions as supermajority vote requirements and blank check preferred stock offer any incremental takeover deterrence, or whether the widespread availability of shadow poison pills (pills that can be quickly adopted without shareholder approval) renders most other provisions irrelevant. In short, there is little consensus among researchers about which, or whether, antitakeover provisions affect a firm’s takeover likelihood in meaningful ways.

Reflecting this ongoing debate, finance researchers have yet to agree on what measures of takeover deterrence are useful in empirical tests. Some researchers use the G-index or E-index, while others use personally-chosen subsets of provisions they regard as most relevant, such as the use of classified boards. Underlying these various attempts to control for firms’ takeover defenses is a lack of agreement about a basic question: Which provisions matter, or matter most? That is, which ones actually decrease a firm’s takeover likelihood?

Our paper, Which Antitakeover Provisions Matter?, recently made publicly available on SSRN, seeks to answer this question. We acknowledge theoretical arguments that some provisions are more important than others. But we base our inferences on data-driven, atheoretic tests that examine the empirical relation between firms’ uses of specific provisions and their takeover likelihoods. Our approach is similar to that in Karpoff, Schonlau, and Wehrly (2017), who show that both the G-index and E-index are negatively related to takeover likelihood after accounting for endogeneity. Karpoff et al.’s (2017) findings imply that the G-index and E-index pick up some aspect of takeover deterrence. But they do not identify which provisions in these indices drive their results and which provisions contribute mostly noise. That is, they do not identify which provisions matter most for takeover likelihoods. The current paper addresses this issue.

To do so, we measure the impact of each provision on firms’ takeover likelihoods. Our tests are complicated by the possibility that firms’ uses of takeover defenses are endogenously related to their takeover likelihoods. For example, firms that deploy certain takeover defenses may do so precisely because they are likely to receive unsolicited takeover bids. Hence, the absence of an empirical correlation between specific provisions and firm independence could be consistent with the idea that some provisions do in fact deter takeovers, but tend to be deployed by firms with high takeover likelihoods. Similarly, a spurious empirical correlation between a provision and takeover likelihood might result from an omitted variable—e.g. a managerial disposition against takeovers—that causes both the use of the provision and a lower takeover likelihood. In such a case, the omitted variable could create the illusion that a particular defense is effective, when instead it is simply endogenous.

To address concerns about endogeneity we adapt the empirical approach in Karpoff et al. (2017) and employed in Amihud, Schmid, and Davidoff Solomon to examine firms’ uses of specific provisions. For each provision, we construct two instruments to identify exogenous variation in a firm’s use of that provision. The instruments rely on predetermined information from non-industry matched geographical and IPO-year-based peer groups. Following guidelines discussed in the literature, we test and confirm that our instruments meet the requirements for strong instruments. The instruments allow for causal estimates of each individual provision’s effect on takeover likelihood despite its potentially endogenous nature. To ensure our inferences are robust we probe our conclusions using specifications that include or exclude the other G-index provisions when estimating the marginal effect of a given provision. We also employ a variety of empirical approaches to account for endogeneity that require different underlying assumptions, including two-stage least squares, limited information maximum likelihood, and recursive bivariate probit models.

Using these models, we find that 11 of the 24 G-index provisions are robustly and negatively related to takeover likelihood, and two provisions are positively related to takeover likelihood. The eleven provisions that negatively relate to takeover likelihood are anti-greenmail provisions, blank check preferred stock, classified boards, director contracts, director indemnification, director liability provisions, directors’ duties provisions, executive severance contracts, fair price provisions, supermajority vote requirements, and unequal voting rights. (Poison pills are notably absent from this list. This is not surprising because all firms to some degree have a shadow pill during our sample period, so our empirical approach lacks the ability to detect the marginal effect of poison pills.) The provisions positively related to takeover likelihood are golden parachutes and restrictions on action by written consent. The fact that these two provisions are positively related to takeover likelihood runs counter to the assumptions behind their widespread use in indices meant to capture firms’ takeover defense. Our results indicate that each of these 13 provisions have incremental effects on takeover likelihood, a result that is inconsistent with arguments that only one or two provisions (e.g., poison pills, classified boards) affect takeover likelihoods.

In addition to the provision-level analysis, we use our empirical approach to compare different subsets of provisions used in previous papers. Researchers have argued that various subsets of provisions best measure a firm’s takeover defenses. To evaluate these claims, we test whether these subsets relate to takeover likelihood in a robust way after controlling for the endogeneity of the provisions. As part of this analysis we construct an index (i.e., the Deterrence-index, or D-index) of takeover defense based on the 13 provisions that are significant in our provision-level analysis. As with the G-index and E-index, the D-index likely overweights some provisions and underweights others. Despite these limitations, our evidence implies that the D-index is a less noisy measure of a firm’s takeover defense. In particular: (i) the D-index is negatively related to takeover likelihoods with or without accounting for endogeneity; (ii) the predictive power of each set of provisions used in previous analyses to measure takeover deterrence, including the G-index and E-index, is attributable to the subset of provisions that overlap with the D-index; and (iii) the D-index is significantly (and negatively) related to firms’ unconditional takeover premiums, whereas the results for the G-index and E-index are mixed.

The results of our paper make six main contributions to the literature. First, we provide empirical evidence that different antitakeover provisions do indeed have different effects on firms’ takeover likelihoods, as has long been theorized by previous researchers. However, and second, we show that only a subset of the 24 G-index provisions relate to takeover likelihoods. Third, two G-index provisions are positively, not negatively, related to takeover likelihood. Fourth, the E-index excludes several provisions that are empirically related to takeover likelihoods and includes one provision that is positively related to takeover likelihood. Fifth, our results provide evidence for several long-running theoretical debates about the impacts of many individual antitakeover provisions. For example, we show that, while classified boards are important, they are but one of several antitakeover provisions that offer takeover defense, while golden parachutes are positively, not negatively, related to takeover likelihood. And sixth, our results show why the G-index and E-index serve as workable but noisy proxies for takeover susceptibility—they combine provisions that do and do not empirically relate to takeover likelihoods, they equally weight provisions that have different marginal effects, and they aggregate together provisions with both positive and negative marginal effects on takeover likelihood.

The full paper is available for download here.

Both comments and trackbacks are currently closed.