07 February 2008

Banks as dangerous neighbourhoods


Tom Keene interviews Nassim Taleb about the deficiencies of Value at Risk models by explaining that empirical models cannot predict Jerome Kerviel's 5 Billion euro loss at SocGen on the basis of existing data, for example Nick Leeson's 860 Million loss at Barings.

The progression can only be explained by evoking nonlinear processes, says Taleb. Taleb suggests that the nonlinearity can be explained by the trend towards concentration in the banking industry (fewer banks, each larger in size) and by increased interdependencies between these banks (eg increased use of interbank loans).

The solution? Smaller banks; and more of them.