Steven Davidoff Solomon is Professor of Law at the University of California at Berkeley School of Law; Byung Hyun Ahn is a Researcher at Dimensional Fund Advisors; and Panos N. Patatoukas is Associate Professor of Finance at the University of California at Berkeley Haas School of Business. This post is based on their recent paper.
Related research from the Program on Corporate Governance includes Why Firms Adopt Antitakeover Arrangements by Lucian A. Bebchuk.
Index constructs and identifiers are regularly utilized in empirical corporate governance research. They are popular. The use of legal changes as plausibly exogenous sources of variation in the economic determinants of corporate governance is also common. State antitakeover laws, particularly business combination laws, are often used as exogenous identifiers to assess corporate governance effects. Despite their popularity, there is a growing body of literature questioning the interpretation of tests that use legal changes for identification.
In Identifying Corporate Governance Effects: The Case of Universal Demand Laws recently posted to SSRN, we contribute to the debate by examining the adoption of Universal Demand (UD) laws, an increasingly popular proxy for exogenous variation in corporate governance mechanisms. UD laws were enacted by 23 states between 1989 and 2005 and require that shareholders make a demand on the board before suing for breach of fiduciary duty or other derivative actions. Because the board can refuse the demand or otherwise prosecute the case, or decline to prosecute, academics have theorized that UD laws decrease the ability of shareholders to litigate and effectively monitor the board. Since UD laws are exogenously imposed by the state, they have the potential to address the issue of endogeneity in the relation between corporate governance and litigation risk.
In a novel paper, Appel (2019) first deployed the enactment of UD laws as a plausibly exogenous source of variation in the use of entrenchment provisions commonly opposed by shareholders. Appel’s empirical investigation zeroes in on variation in the widely-used entrenchment index (E-Index), which captures the sum of provisions restricting shareholder voting power and antitakeover provisions (Bebchuk et al. 2009). The key finding is that the enactment of UD laws across adopting states is associated with a significant increase in the E-Index. Prior work interprets this finding as prima facie evidence of a causal link between shareholder litigation rights and corporate governance. A fast-growing stream of studies in corporate finance and accounting relies on the adoption of UD laws to identify cause-and-effect links between management entrenchment and various firm outcomes. The common thread across these studies is that UD laws had a direct effect on management entrenchment.
If the direct link between the adoption of UD laws and management entrenchment is broken, prior evidence on the cause-and-effect link between UD laws and various firm outcomes becomes questionable.
Given the ambiguous a priori effects, we raise the possibility that prior evidence on the association between UD laws and the use of entrenchment provisions is confounded by measurement and identification issues. As an alternative explanation, we explore the limitations in the ISS database and the resulting misclassification of pre-event changes in the E-Index as post-event changes. Our paper addresses these issues head-on. To identify the effect of the adoption of UD laws on the use of entrenchment provisions, we implement a standard two-group, two-period (22) difference-in-differences (DID) separately for each UD law adopting state. Our research design compares treated and control firms in terms of the pre-post change in their E-Index before and after the adoption of UD laws separately for each adopting state.
The treated group includes firms incorporated in UD law adopting states with coverage in the Institutional Shareholder Services (ISS) legacy database between 1990 and 2006. The control group includes firms with ISS coverage incorporated in states that never adopted UD laws. With respect to the pre-post comparisons, we ensure that our design hews closely to the timeline of the ISS survey release dates. Specifically, we collect data on the use of entrenchment provisions and compute the change in the E-Index between the last ISS survey before the UD law effective date and the first ISS survey after the UD law effective date. Our design further ensures that the control group of never treated firms is not contaminated by comparisons of later versus earlier treated firms. Our analysis reveals that between 1990 and 2006 there are 110 unique firms incorporated in UD law adopting states and have coverage between consecutive ISS surveys centered on the UD law effective date of each adopting state. Importantly, out of the 110 treated firms, we find that only 20 cases appear to experience a pre-post increase in their E-Index, with 10 out of those 20 cases incorporated in Pennsylvania.
Next, we probe the adoption of individual entrenchment provisions across the 20 treated firms that seemingly experienced an increase in their E-index after the adoption of UD laws. The ISS survey data suggest that these 20 treated firms collectively adopted a total of 23 management entrenchment provisions, with poison pill and golden parachute antitakeover provisions accounting for 21 out of the 23 provisions. However, an important limitation of the ISS legacy database is that there is significant lag between consecutive ISS surveys that surround the UD law effective date in each adopting state. In addition, the ISS legacy database does not provide information about the exact adoption dates of individual entrenchment provisions across firms. Put differently, while one can measure changes in the E-Index between consecutive ISS surveys, the ISS legacy database does not allow for the identification of the exact timing of such changes.
To overcome this limitation, we hand-collect information on the exact adoption dates of individual entrenchment provisions across treated firms. Our case-by-case investigation sheds new light on measurement and identification issues in empirical research using the ISS legacy survey data. Recall that across the 20 treated firms with a seeming pre-post increase in their E-Index, the legacy database identifies 23 entrenchment provisions adopted between consecutive ISS surveys. Strikingly, we find that 16 out of the 23 entrenchment provisions were adopted at least four quarters prior to the enactment of UD laws. The evidence highlights the prevalence of the misclassification of pre-event changes in the use of entrenchment provisions as post-event changes in the E-index. In fact, the vast majority of changes in the E-Index among affected firms in UD law adopting states actually occurred well before the enactment of UD laws.
Since the vast majority of changes in the use entrenchment provisions are in fact announced before, often well before, the enactment of UD laws, we conclude that changes in the E-Index across treated firms cannot be causally attributed to UD laws. Next, we probe the fundamental reasons behind the decision to adopt entrenchment provisions. With this objective in mind, we focus on the group of 20 affected firms that adopt entrenchment provisions. For these cases, we measure the cumulative stock return performance leading to the date of adoption of entrenchment provisions. We document that this group underperforms the benchmark portfolio by as much as 38% in the two years leading to the change in their E-Index. The evidence highlights that the use of entrenchment provisions, and most notably antitakeover tactics, is endogenously determined. Declining firms in UD law adopting states adopt poison pills and golden parachutes as antitakeover provisions only after substantial long-term drops in value. The strong link between the adoption of entrenchment provisions and past stock return performance casts doubt on the alternative possibility that management chooses to become more entrenched in anticipation of the enactment of UD laws and a potential decrease in future derivative litigation risk.
Overall, our granular investigation of the use of entrenchment provisions among affected firms shows that the link between UD laws and corporate governance is influenced by a small number of firms adopting antitakeover provisions after substantial long-term drops in value. While prior studies often rely on a fundamental premise that the protection from suit the UD laws might bring will engender a change in corporate governance, we conclude that the connection between corporate governance and UD laws is still questionable at best. The issue of misclassification of pre-adoption changes in the use of entrenchment provisions as post-adoption changes in the E-Index has not been explored in prior UD law research. The misclassification issue has broader implications for studies that zero in on the identification of the timing of changes in management entrenchment using the ISS legacy database and adds to research highlighting the difficulty of coding and measuring corporate governance.
In short, we believe there should be a healthy skepticism of corporate governance identifiers without a firm theoretical channel identified. In addition, studies should beware of small sample bias as well as the inherent limitations of current corporate governance indexes.
The complete paper is available for download here.