Stress-Testing

Anticipating what could go wrong in a crisis

The EMA system includes the ability to stress test a portfolio to estimate its performance should particular events occur. The analysis can either simply re-price all assets based on a shift in the value of an underlying instrument or it can utilise the factor model to extrapolate from a shift in one instrument to the likely consequences for all other assets.

Stress-testing has become more common, with regulators insisting on the testing of specific or user-determined scenarios, following the 2008 crisis when markets performed outside the range of outcomes that appeared likely based on historic behaviour. The stress-test is designed to capture the effects of “Black-Swans” or “6 standard-deviation events” – the crises that appear so unlikely that ordinarily the risk of them happening is dismissed.

Stress-testing can also be used in “Scenario Analysis” where a non-crisis event is simulated and the system indicates each assets likely behaviour should it occur. This can be useful for a manager with a top-down view who is seeking assistance in considering the best assets to hold given the macro forecast.

The system includes, by default, a set of defined stress events based on historical events or standard scenarios such as a decline in the US Treasury bond market or in Global equity markets. In addition, users may define their own stress events where a shift can be expressed in nominal terms, as a standard deviation of the returns or, in the case of a yield curve, as a multiple of the rate at each point on the curve.

To make it possible to explore worst-case scenarios, the system can be set to consider a matrix of shifts, where a range of magnitudes of shift are tested for each instrument comprising the stress scenario. This allows for “Reverse Stress Testing” where it is possible to see the set of macro events that would be most harmful to the value of a given portfolio.

 

Stress testing features
Types of calculation Independent – reprice based on instrument shift Factor model – use correlation model to estimate implied price moves
Types of stress Historic (e.g. 2008) Scenario (e.g. Global inflation increase) Standard (e.g. all major equity markets off 10%)
Types of shift Nominal (e.g. -10%) Standard deviation (e.g. 3) Multiple (e.g. all points on yield curve increase by 1.2x)
Matrix All combinations of instrument shifts within range (e.g. (i) SP500 -5%, WTI +10%, (ii) SP500 -10%, WTI +10%, (iii) SP500 -15%, WTI +5%, etc)