|
|
By Peter Urbani, CIO, Infiniti Capital:
An obscure tract by a University of Adelaide Statistics Professor, Edmund Cornish is today among the leading candidates for improving risk management.
The 1937 paper by Edmund Alfred Cornish (1909 - 1973) and Sir Ronald Fisher* provides the basis for the Cornish-Fisher expansion by which the impact of higher statistical moments such as skewness and kurtosis (3rd and 4th statistical moments) can be added to the normal distribution.
This is important in the measurement of risk because these higher moments are primarily what is responsible for the so called 'fat tails' of returns. These cause large losses to be both more frequent and more severe than predicted by the normal distribution which considers only the first two moments (Mean and Standard Deviation). The Gaussian or Normal distribution underpins all of probability theory and the 'assumption of normality' is deeply embedded in most finance theory including option pricing models and the widely used Value at Risk (VaR) metric by which banks determine how much capital they need to hold in reserve against potential losses.
As we have seen from the recent credit crisis and a spate of bank failures in the US and elsewhere (72 US banks have failed so far this year), most banks were not holding sufficient capital to cover their losses in the recent crisis. In fact it has been estimated that in aggregate they were holding only half as much in reserve as they actu...................... To view our full article Click here
|
|