July 09, 2016
Sir, I refer to Robert Armstrong’s lunch with Philip Tetlock, ‘It doesn’t matter how smart you are’, July 9.
How I would have loved being at that table and be able to ask them:
Gentlemen, where do you think the next major bank crisis resulting from excessive exposures to something really bad is going to happen, between something that was perceived as safe when incorporated to bank’s balance sheets, or between something that was ex-ante perceived as risky?
And applying simple common sense, and looking at empirical evidence, I would absolutely forecast the first, that what’s perceived ex-ante as safe is, ex-post, the much riskier.
And I ask this because our current regulators, with their risk weighted capital requirements for banks, forecast that what is ex-ante perceived as risky, is ex-post, what is truly dangerous.
In my mind they never heard of Voltaire’s “May God defend me from my friends [AAA rated]: I can defend myself from my enemies [BB- rated]”
And the biggest problem now is that, though the regulators were clearly proven wrong in 2007-08, they do not admit their mistake and they keep on forecasting the same.
Sir, if artificial intelligence is to help us, we must keep it free of weak human egos.