"Weak" Governance May Be Optimal Governance: A Discussion of "Corporate Governance and Backdating of Executive Stock Options"*
讨论Collins、Gong和Li（2009）关于公司治理与高管股票期权回溯的论文，指出其将期权回溯归因于弱治理结构，但质疑统计关联的解释力，对实证公司治理研究提出根本性批评。
In this discussion, I provide some big picture thoughts on the Collins, Gong and Li (2009) paper (CGL hereafter) entitled: Corporate Governance and Backdating of Executive Stock Options. CGL utilize an estimation algorithm to generate a large sample of potential incidences of grant-level backdating behavior. The conceptual premise of the paper is that the backdating of CEO stock option grants is a direct consequence of “weak ” corporate governance structures. The authors make directional predictions about the relation between firms ’ individual governance structures and the probability that stock option backdating occurs. Each governance structure for a firm is represented by a numerical metric where higher values are interpreted as weaker governance. For example, governance is posited to get weaker as the proportion of inside or gray directors increases. The authors then predict the probability of backdating to be an increasing function of the metric representing each individual governance structure. CGL adds to the stream of papers that empirically examine the backdating issue (e.g., Heron and Lie (2007), Bebchuk, Grinstein and Peyer (2007), and Bizjak et al., (2006)). In some sense, CGL is the culmination of this research line. CGL replicates a number of results spread across the extant literature, and add some twists and turns of their own. The authors do a nice job in the paper of isolating their unique contributions, so I need not repeat any of that here. Overall, this is a very careful piece of empirical research, and I have no intention of nit-picking the empirical analysis. Instead, I offer a critique of the paper that is pertinent to the entire literature on backdating and to empirical corporate governance research in general. In particular, I consider fundamental problems in interpreting statistical associations between corporate behavior and measures of individual corporate governance structures. A large body of corporate governance research, including CGL, regress measures of corporate behavior on metrics of corporate governance attributes. But this design raises a series