The need for simple financial regulations

As argued previously (here and there), financial sectors need simple regulations, not more complicated ones. This is being voiced more frequently than ever. Nicholas Brady calls for a new simplicity agenda, in the FT:

Regulators need a clear “bright line” that they can apply to bank activities. The aim should be to permit innovation, and prudent risk taking, while also creating less varied and complex boundaries that banks cannot cross and that everyone can understand. The new simplicity should establish a clear ability to determine when to say yes, and when to say no; and the meaning of “no” should be unambiguous.The debate should shift to focus on the total leverage permitted in the bank’s books – that is the “bright line”. Banks should be permitted to devise their own strategies and use trading as they see fit, but they should be restricted from taking positions that use leverage of more than “X-to-1”. That may limit the upside of their operations, but at the same time it will limit the downside for taxpayers. It also puts responsibility for operational decisions where they belong, in the hands of the bankers themselves. What should be the boundary? Will it take additional time to design? Yes, but it will be worth it. For a change the public will both understand and agree with it.

The case of "less is better" is gaining more weight against the traditional paradigm of regulation. Some are indeed pushing for more and different way of data gathering about risks and quantities. But the last financial crisis reminded us of the difference between risks (quantifiable) and uncertainty (which is not). It took back from the shelf Knight, Keynes and Minsky and  in a speech at the Fed's Jackson Hole conference, Mr. Haldane from the Bank of England:

The degree of complexity also raises serious questions about the robustness of the regulatory framework given its degree of over-parameterisation. This million-dimension parameter set is based on the in-sample statistical fit of models drawn from short historical samples. If previous studies tell us it may take 250 years of data for a complex asset pricing model to beat a simple one, it is difficult to imagine how long a sample would be needed to justify a million-digit parameter set.