Cutting Edge introduction: risky contributions

Value-at-risk is usually calculated via Monte Carlo simulation, making it difficult to see the contributions from different risks. But in some circumstances approximate formulas can be derived that greatly save computing power – and explicitly show the sources of risk. Laurie Carver introduces this month’s technical articles

The calculation of risk measures such as value-at-risk – which continues to be widely used despite its shortcomings – is vital to risk management. Quants tend to calculate them in numerical fashion, usually via Monte Carlo simulation, because the functions themselves tend to resist direct formulas. For instance, VAR is a quantile of the underlying returns distribution, which is hard to write down directly.

The exception comes when sensitivities are used as a starting place – with a risk measure