Network Effects in Corporate Governance

Sarath Sanga is Assistant Professor of Law at the Northwestern University Pritzker School of Law. This post is based on his recent paper and is part of the Delaware law series; links to other posts in the series are available here. Related research from the Program on Corporate Governance includes Firms’ Decisions Where to Incorporate by Lucian Bebchuk and Alma Cohen; The Market for Corporate Law by Lucian Bebchuk, Oren Bar-Gill and Michal Barzuza; and Delaware Law as Lingua Franca: Evidence from VC-Backed Startups by Jesse Fried, Brian J. Broughman, and Darian M. Ibrahim (discussed on the Forum here).

Why Does Everyone Incorporate in Delaware?

There are two canonical explanations:

(1) Legal quality. Firms are influenced by the intrinsic qualities of Delaware’s legal system. By some reasonable measure, its statutes, common law, and expert courts are “the best.”

(2) Network effects. Firms are influenced by each other’s corporate governance decisions. Why? Perhaps they interpret these decisions as proof of Delaware’s quality. Or perhaps they only want to follow the trend. In either case, this is a self-perpetuating rationale: Everyone goes to Delaware because everyone else is already there.

In a new paper, I analyze the incorporation histories of over 22,000 public companies from 1930 to 2010. I show that network effects were the principal force behind Delaware’s ascendance.

I find that Delaware’s share of public companies was virtually constant from 1930 to 1985. Then came a watershed event: the case of Smith v. Van Gorkom in 1985 and the subsequent passage of section 102(b)(7) in 1986. Since then, Delaware’s share has more than doubled.

As is well known, the Van Gorkom decision shocked the business and legal community when it imposed personal monetary liability on a director for breach of the duty of care. The Delaware legislature responded swiftly with section 102(b)(7), enabling a corporation ex ante to waive such liability and thus avoid outcomes like Van Gorkom.

I argue that we can use this singular event to distinguish between the two canonical theories, that is, between the influence of changes in the law (legal effects) versus changes in other firms’ decisions (network effects).

To see this, suppose—just for the moment—that corporate law did not change again after 102(b)(7). Now consider the incorporation decisions of new public companies. After 102(b)(7), Delaware law is suddenly more attractive for the litigation-wary director. Thus, we would expect Delaware’s share of new firms to suddenly jump after 102(b)(7).

But if new firms were also influenced by changes in other firms’ decisions, then we would further expect that initial jump to be followed by another jump—and another, and another. The intuition is that the first cohort of new firms responds only to the legal shock. The second cohort of new firms, however, responds to the legal shock and the initial jump in the first cohort’s behavior. The third cohort of new firms responds to the legal shock and the initial jump and the second cohort’s response to the initial jump—and so on. The result is a cascading effect in which each cohort successively reacts to the previous cohort’s reaction.

This is precisely what I find in the data. I estimate that after 102(b)(7), Delaware’s share of new public companies immediately jumped from 30 to 43 percent. It then continued to rise, quickly at first, then slower in later years. By 1990 it reached 60 percent. By 2000 it stabilized at 74 percent. Thus, the change in legal quality brought by 102(b)(7) is responsible for the initial rise from 30 to 43 percent, while reactions to other firms’ reactions (network effects) are responsible for the subsequent rise from 43 to 74 percent. Network effects, therefore, are responsible for 70 percent of the total change.

However—the argument above was premised on the assumption that corporate law did not change again after 102(b)(7). Strictly speaking, this is false. Delaware law has changed countless times since. Could it be that those subsequent legal changes were responsible for the cascading effect we observe?

I conclude no. What matters is not whether the law has changed, but whether the law has changed in ways that significantly affect the decision to incorporate in Delaware. We can infer that it has not by studying reincorporation rates. Changes in these rates reflect changes in the extent to which existing public companies are attracted to Delaware. As it turns out, the reincorporation rate has been virtually constant for the last 50 years—with one big exception. In 1987 and 1988, the two years following 102(b)(7), the rate of reincorporation spiked more than five-fold and over 300 public companies moved to Delaware. By 1989, reincorporations plummeted back to pre-102(b)(7) levels, and they have hardly changed since.

This pattern suggests that the seemingly unrealistic assumption—that corporate law has not changed since 1986—is approximately true, at least insofar as incorporation decisions are concerned. The relative attraction of Delaware’s laws has been roughly constant for the last half-century, save for one massive shock in 1986. (The full story is a bit more nuanced, and the paper offers a formal mathematical model to account for these subtleties. Suffice it to say that the account given here is a reasonable approximation.)

Finally, the stock market returns of Delaware companies offer evidence that Delaware’s vast network of firms is itself a source of value. I construct a hypothetical zero-investment portfolio that buys Delaware firms and sells short non-Delaware firms. I then analyze the returns of this portfolio from 1930 to 2010. Unsurprisingly, the returns are typically zero.

However, there were three brief periods of significant positive abnormal returns. The first two periods were during the Great Depression and just after World War II. The third and final period coincides precisely with the years after 102(b)(7) passed—i.e., the years in which the Delaware network grew most. These returns were 3.4 percent per year from June 1986 (when 102(b)(7) passed) to May 1990. Further, since most other states passed similar statutes within a few years, we can use firms incorporated in those states as a control group to test the hypothesis that 102(b)(7) is itself responsible for the abnormal returns. The data strongly reject this hypothesis.

Why would Delaware corporate law become more valuable as more firms adopt it? The existing literature offers several theories. [1] For example, when a firm incorporates in Delaware, it saves directors the burden of learning a new set of duties as they move from one Delaware firm to another, thus encouraging director mobility and information spillovers among firms in the Delaware network. Incorporating in Delaware also grants the firm access to a global network of intermediary services that are well-versed in its standards such as international law firms, accounting firms, and investment banks. Moreover, firms themselves may directly increase legal quality via litigation and lobbying, since frequent litigation resolves legal uncertainty even for non-litigant firms, while lobbying benefits can similarly spill over to non-lobbying firms. Indeed, for an example of the latter, one need look no further than 102(b)(7).

The complete paper is available here.


1Prominent examples include Klausner (1995), Kahan and Klausner (1997), Kamar (1998), Coffee (1998), Bebchuk and Roe (1999), Gilson (2001), Hansmann and Kraakman (2001), Kahan and Kamar (2002), Daines (2002), Bar-Gill, Barzuza and Bebchuk (2006), Hadfield and Talley (2006), and Talley (2015).(go back)

Both comments and trackbacks are currently closed.

One Comment

  1. John Baker
    Posted Wednesday, January 17, 2018 at 12:25 pm | Permalink

    Very interesting. I think that the analysis should also take into account the passage of the Delaware antitakeover statute, section 203 of the Delaware General Corporation Law, in 1988. The statute received wide publicity and was a factor in the decision of a number of corporations to reincorporate in Delaware.
    Consistent with your thesis, the statute was by no means the only or the strongest antitakeover statute. It was, however, the most discussed.