Under Hypothesis III, all of the results in the previous section would be driven by omitted-variable bias. Since governance provisions were certainly not adopted randomly, it is plausible that differences in industry, S&P 500 inclusion, institutional ownership, or other firm characteristics could be correlated both with G and with abnormal returns. Under this hypothesis, governance provisions could be completely innocuous, with no influ
ence either on managerial power or on agency costs.