The sample was constructed in the following way. First, we dropped all company- years that do not report either executive or non-executive boards. Second, we removed all firms that report either negative equity or negative total assets. At this point the data consisted of about 57,000 director-company-year observations pertaining to 8,506 firm-years. Third, to address the issue of firms in severe financial distress, we have dropped companies that report ROE or ROS less than -1. Fourth, to reduce the effect of outliers we dropped 1% observations from the left and right tails of the distribution of performance.
