Risk management for non-normality

CFA Level III
#1

Reading bonds rights now and apparently shortfall risk is good risk management tool for bonds with embedded options. I thought shortfall risk measures the std dev of downside so doesn’t that assume a normal distribution? In general which metrics are good for non normal? VAR is out, std dev, var etc…

#2

AC123 Wrote: ------------------------------------------------------- > Reading bonds rights now and apparently shortfall > risk is good risk management tool for bonds with > embedded options. > > I thought shortfall risk measures the std dev of > downside so doesn’t that assume a normal > distribution? What you described is semivariance. Shortfall risk is the probability with that. > In general which metrics are good for non normal? > VAR is out, std dev, var etc… Good question. It seems the LOS only says we are to critique all of them, never suggesting the best candidate for bond risk measurement :slight_smile: - sticky

#3

Right you are, shortfall risk is a subset of down side risk measure along with semi var and safety first. What other risk measures apart from shortfall risk can we use w/ options?

#4

Wouldn’t Historical Method help you find VaR because it doesn’t depend on normality of returns? Also Monte Carlo simulation does not require normal distribution, so you can calculate VaR with that too.

#5

abacus is correct, both historical and Monte Carlo VAR calculation will be fine methods to deal with non-normality problem. Both bethods have their flaws, for instance historical method solely relies on historical data, while Monte Carlo maybe costly and computationally chalenging to run. But as far as non-normal distributions are concerned, they would do the job.