
volatility of the “real” results of the business units.
6
Since it is not always possible to scale
monthly return volatility into annual volatility in a simple manner, the annual earnings-
at-risk fi gure may be derived by numerically simulating a series of monthly shocks, and
autocorrelation — provided it exists — in monthly returns could simply be taken into
account through the same simulation procedure.
A third problem relates to the potential need to rescale earnings series and/or trans-
form them into some sort of “returns.” First of all, if the size of the business unit has
substantially changed through time, higher or lower earn