In our paper, “Thirty Years of Corporate Governance: Firm Valuation & Stock Returns”, which we recently presented at the Seminar in Law, Economics, and Organization here at Harvard Law School, we introduce a comprehensive corporate governance database starting in 1978 and ending in 1989, which tracks for a sample of approximately 1,000 unique firms whether these firms had any of the 24 corporate governance variables that constitute the G-Index of Gompers, Ishii and Metrick (2003). The computation of this index for the 1990-2006 period is based on data compiled by the IRRC (Investor Responsibility Research Center). The G-Index is a composite of the twenty-four variables, adding one point if any of the provisions is present, where a higher score indicates more restrictions on shareholder rights or a greater number of anti-takeover measures. The E-Index of Bebchuk, Cohen, and Ferrell (2009) is based on six of the twenty-four G-Index provisions. By combining our dataset with the IRRC database which covers the 1990-2006 time period, we obtain comprehensive corporate governance data for the 1978-2006 time period. The importance of having data for the 1978 – 1989 period is underscored by the fact that this period is characterized by widespread corporate governance changes, while after 1990 such changes largely cease. Further, four out of the six E-Index provisions (supermajority merger, classified board, poison pill and golden parachute) experienced dramatic increases in their incidence during the 1978-1989 period, with their incidences remaining relatively stable thereafter. In addition to the introduction of our database, our paper addresses two questions: what is the relationship between governance and firm valuation and what is the relationship between governance and abnormal stock returns over the 1978-2006 time period.
Turning to the central issue of the relationship between governance and firm valuation, we find a robust statistically significant negative association between poor governance and firm valuation over the 1978-2006 period. In particular, the inclusion of firm fixed effects in pooled panel regressions mitigates the endogeneity of firms adopting governance provisions depending on their heterogeneous circumstances. Using both firm and year fixed effects, we document a robust negative and economically meaningful association between the G-Index and Tobin’s Q. This finding survives various robustness checks. The economic magnitude of the association seems meaningful. For example, over the full time period and using firm and year fixed effects, the coefficient of the G-Index equals -0.011 implies that a one standard deviation increase of the G-Index (3.0) is associated with a decrease in firm value of about 3.3%. We find, however, no evidence in support of the “reverse causation” explanation for this negative association in the 1978-1989 period, i.e. that firms with lower firm value tended to adopt more G- and E-Index provisions. In fact, we find that the higher valued firms tended to adopt more provisions, although this relationship disappears once firm and year fixed effects are included. The “reverse causation” may play a (economically very minor) role in explaining changes in firms’ corporate governance in the 1990-2006 time period.
Turning to the relationship between firm valuation and abnormal stock returns, Gompers, Ishii and Metrick (2003) document that firms with higher (lower) G-Index scores have lower (higher) subsequent stock returns. Our longer time period, and the time variation in corporate governance arrangements, enables us to make a number of findings that bear on the literature that developed from this finding. We document that for the full 1978-2007 time period, whether using value-weighted or equally-weighted portfolios, that there are positive, strongly statistically significant positive abnormal returns (using the Fama-French-Cahart four-factor model) associated with going long good corporate governance firms and shorting those with poor governance (whether proxying the quality of corporate governance by the G- or E-Index). Second, we find that abnormal returns for equally-weighted portfolios over the 1978-2007 period is robust to industry-adjusting. However, the abnormal returns of the value-weighted portfolios are not robust to industry-adjusting. Third, our analysis of returns suggest that governance seems to matter most for smaller capitalization stocks and that the association between governance and abnormal returns generally appears to decline over our time period. We interpret our governance-related abnormal return findings as consistent with ‘learning,’ i.e., investors learned gradually over time the importance of good governance, which is reflected in the fact that abnormal returns were largest in the beginning of our time period and then generally declined thereafter.
The full paper is available for download here.