Risk measurement involves estimating some functional of the
distribution of loss.
some risk measures, such as value at risk and tail
conditional expectation, are
not means of a distribution from which one can sample. This
calls for nested
simulation, in which risk factors are sampled at an outer
level of simulation,
while the inner level of simulation provides estimates of
loss given each
realization of the risk factors. In our examples, the outer
level simulates
tomorrow's stock prices, and the inner level estimates the
loss of a portfolio of
stock options given tomorrow's stock prices. We present a
general method for
providing a confidence interval for the risk measurement
given statistical error
at two levels of simulation. This method could require a
large computational
budget, so we discuss efficient procedures for providing a
confidence interval
and point estimates for tail conditional expectation.