August 24, 2014

Bank regulators should stop profiling risk and use predictive statistics instead.

Sir, Gillian Tett brings up an extremely interesting question. What has predictive statistics on crime, which might indicate more crime possibilities in black areas, have to do with discrimination? “Mapping crime – or stirring hate” August 23. I would apply those notions to current banks regulations.

Regulators should apply risk-neutrality, meaning stop profiling bank assets using risk-weights based on perceived risk, which in essence is highly discriminatory and creates distortions in the allocation of credit; and instead use predictive statistics about when banks really get into troubles. Would they do so, they would soon discover that bank lending to “the risky” requires much less supervision than bank lending to what is perceived as safe and profitable. 

The predictive model could be based on quite simple algorithms… like what bank exposures are growing the fastest, in real time. At this time it would clearly indicate that, in Europe, regulators would have to urgently send out a squad to patrol the area of loans to infallible sovereigns.