I continue to receive reactions to this new paper from Glass Lewis entitled “AI and the Fiduciary Test: A Guide for Institutional Investors in Evaluating AI Proxy Voting Solutions” – and I received a number of reactions to the last blog I posted in this series – including this reaction:
“Nicolaas Koster of Proxywise AI cites votes on over-boarded directors as a perfect use case for AI. But his underlying assumption is that investors are just mechanically voting against all directors who violate their overboarding policies. And that’s not necessarily the case.
For example, investors, and the proxy advisors, will make exceptions for temporary overboarding (i.e. when a director joins a new board a few weeks prior to stepping down from an old board). Sometimes that’s disclosed in a proxy statement, but sometimes we have to look to the proxy (or an 8-K) of the company whose board the director is exiting. At the same time, investors generally care about a director’s service on non-US boards, which US companies are not required to disclose in their proxies, in addition to service on US boards. Unless the AI model has comprehensive global data, it will miss such nuances.
Most other ‘routine’ governance votes are similarly not as straightforward as they first appear. For example, most investors and proxy advisors have policies on director attendance at board and committee meetings. But it’s not a matter of mechanically voting against any director who fails to clear the 75% threshold: investors, and certainly the proxy advisors, look for disclosure of a reason for the poor attendance.
If the reason is considered valid (e.g. a medical issue or family emergency; or service on a presidential commission), investors will give the director a pass. It may be that AI models can be trained to distinguish between valid reasons and those that are not, but only if the people designing those models are attuned to the need to do so.
And of course, there are many proposals on meeting ballots that are vastly more numerous than contested elections or novel shareholder proposals, but also vastly more complex than evaluating director attendance. Say-on-pay is the obvious example, but also reincorporation proposals, bundled charter amendments, financing proposals and many others.
There are usually genuine pros and cons in all of these cases and investors who take their own fiduciary duties seriously can’t simply rely on the board’s stated rationale, especially when the directors’ interests can conflict with those of shareholders.
So I disagree that most ballot items ‘can be decided deterministically from disclosed facts and a client’s stated policy.'”
