V.D. Discussion
The evidence in subsections V.A, V.B, and V.C must be in terpreted with caution. Since this is an experiment without random assignment, no analysis of causality can be conclusive. The main problem is the possibility that some unobserved character istic is correlated with G and is also the main cause of abnormal returns. This type of omitted-variable bias could be something prosaic, such as imperfect industry adjustments or model misspecification, or something more difficult to quantify, such as a partially unobservable or immeasurable "corporate culture." Under the latter explanation, management behavior would be constrained by cultural norms within the firm, and democracy and dictatorship would be a persistent feature of a corporate culture; G would be a symptom, but not a cause, of this culture. In this case, all the results of the paper could be explained if investors mispriced culture in 1990, just as they appear to have mispriced its proxy, G. The policy impact of reducing G would be nonexistent unless it affected the culture of managerial power that was the true driver of poor performance.