Although empirical scholarship dominates the field of law and finance, much of it shares a common vulnerability: an abiding faith in the accuracy and integrity of a small, specialized collection of corporate governance data. In this paper, we unveil a novel collection of three decades’ worth of corporate charters for thousands of public companies, which shows that this faith is misplaced.

We make three principal contributions to the literature. First, we label our corpus for a variety of firm- and state-level governance features. Doing so reveals significant infirmities within the most well-known corporate governance datasets, including an error rate exceeding eighty percent in the G-Index, the most widely used proxy for “good governance” in law and finance. Correcting these errors substantially weakens one of the most well-known results in law and finance, which associates good governance with higher investment returns. Second, we make our corpus freely available to others, in hope of providing a long-overdue resource for traditional scholars as well as those exploring new frontiers in corporate governance, ranging from machine learning to stakeholder governance to the effects of common ownership. Third, and more broadly, our analysis exposes twin cautionary tales about the critical role of lawyers in empirical research, and the dubious practice of throttling public access to public records.

Citation
Jens Frankenreiter et al., Cleaning Corporate Governance, 170 University of Pennsylvania Law Review, 1–70 (2021).