Stress Testing

CFA Level III
#1

Hi,

Does anyone have any good advice how to remember the differences between scenario analysis and stressing models.

Many of them are so similar that I don’t understand how to distinquish between them during the exam.

For example:

Hypothetical events: events that have never happened in the markets or market outcomes to which we attach a small probability

Worst-case scenario analysis: A stress test in which we examine the worst case that we actually expect to occur.

Like honestly, what is the difference between those two?!

#2

stress testing is a more extreme scenario

#3

what is it exactly that you’re saying?

Stress testing is an umbrella term under which scenario analysis and stressing models fall. Straight out of curriculum:

“Two broad approaches exist in stress testing: scenario analysis and stressing models.”

#4

Scenario can be good also. Stress test is just negative.

Don’t remember seeing a question where I had to differentiate.

#5

Think of a scenario as a complete picture/set of factors (.com, financial crisis, Euro crisis, etc.). Stress testing gets more granular, stressing individual factors. You can stress factors in scenario analysis, by pushing any given factor further.

#6

personally, in my opinion i only see them asking something like recommend one or two other stress test methods to support VaR and explain an advantage.

  1. actual extreme events - use actual events in the past to replicate how the portfolio would perform under the same conditions which can be added to the model to increase the probabilistic outcome based on what the model is predicting or timeframe involved/period of development.

  2. Hypothetical events - blah blah blah like above

#7

Agree, with possible “two colleagues are discussing the correct approach to stressing a model”, they disagree, who’s correct and if not correct explain why.

This setup could incorporate economic forecasting models, ERP, asset allocation, etc. Example, stressing Corp A bond spreads, how does it change markets relative value, which model does it affect. Or stressing the standard deviation of a segmented market, how does it change risk premiums?

Testing multiple concepts in a single question is their objective. I could see a table with results of different stressed factors and you have to discern the impact, model affected, appropriateness, the result - all in 4 min :wink:

#8

I’d be less concerned about stress testing and more concerned about test stressing.

(It’s one of those mornings. :wink:)