Women on boards and corporate social irresponsibility: Evidence from a Granger style reverse causality minimisation procedure

We study the relationship between gender diversity on boards and corporate social irresponsibility (CSI). We hypothesize a bi-directional causality. Firms exposed to CSI incidents are likely to increase their board gender diversity for reputational purposes. At the same time, board gender diversity diversifies board skills and networks, thereby building board capital and enhancing its monitoring function which in turn should result in the ability to reduce CSI incidents. In econometric terms, this relationship is plagued with a reverse causality issue. To address this, we propose a Granger-style reverse causality minimisation procedure. This procedure involves three steps. Firstly, we regress board diversity (BD) on lagged CSI to separate diversity into two components, one driven by CSI (BDDCS) and another unrelated to CSI (BDUCS), with the latter being the sum of the intercept and the disturbance term. Secondly, we confirm that BDUCS experiences a near-zero correlation to CSI (0.01 compared to 0.26 for BD and CSI), and that a Granger causality F-test for CSI affecting BDUCS is clearly insignificant. Thirdly, we regress CSI on lagged BDUCS, lagged CSI and its interaction term. Applying our Granger-style reverse causality minimisation procedure to 2,880 U.S. firms between 2007 and 2016, we find CSI to increase board diversity. Potentially more importantly, we also find that boards with higher diversity, for reasons other than CSI, were significantly better than their lower diversity counterparts in reducing CSI incidents once encountering them. This effect is economically stronger for diversity unrelated to CSI than for overall diversity.