• We are pleased to announce that the winner of our Feedback Prize Draw for the Winter 2024-25 session and winning £150 of gift vouchers is Zhao Liang Tay. Congratulations to Zhao Liang. If you fancy winning £150 worth of gift vouchers (from a major UK store) for the Summer 2025 exam sitting for just a few minutes of your time throughout the session, please see our website at https://www.acted.co.uk/further-info.html?pat=feedback#feedback-prize for more information on how you can make sure your name is included in the draw at the end of the session.
  • Please be advised that the SP1, SP5 and SP7 X1 deadline is the 14th July and not the 17th June as first stated. Please accept out apologies for any confusion caused.

Stress, Scenario, and Sensitivity Testing

E

E-Law

Member
Hello,

I don't understand the ST7 definitions for these tests.

What is the difference between a parameter (stress, scenario) and an assumption (sensitivity)?

And is there anything about stress tests that don't make them also scenario tests?
 
The clue is in the names and can be useful in remembering the descriptions:

A sensitivity test is used to identify which variables have big individual influence on output with just small changes in each variable.

Stress tests on the other hand are used to identify the effect of large changes in a variable, i.e. stressing them.

Finally scenario tests look at the effect of changing multiple variables to create a particular scenario (because stress tests are not so applicable to the real world as variables tend not be stressed in isolation but together as they are correlated).

The first two are single variable tests (small movement, large movement), whereas the third is a multi variable test.
 
Just to add:

Stress tests are normally about extreme values (ie when things are under stress).

With scenario tests the multiple variables are normally varied in a related way.
 
Thank you for your replies, maz and Darren.

My first question wasn't about what they're for, but what is the object you're varying because the definitions use the words "assumption" and "parameter". It isn't obvious to me if there is a difference. From both replies it looks like there is no difference.



My next question was about why is there a separate word for "stress test". The number of parameters to vary seems like a very arbitrary and specific thing to make a distinction for.

We don't have different terms for sandwiches based on how many fillings they have - they're all just sandwiches. Similarly, why bother with the word "stress test" when stress tests are effectively scenario tests?

I'm guessing that there is some reason why the word "stress test" exists because from what's posted here it seems pretty pointless. After all, what's the value to the insurer for finding out the consequence of changing a single parameter a lot? In reality parameters move together due to correlations - hence scenario tests exist. What am I missing?
 
A parameter is a placeholder for an assumption. So you may have inflation as a parameter, and assume a value of 3% for that parameter.

Therefore for a stress test you are not testing your assumption of 3%, but the effect a particularly large assumption would have on the output.

While for a sensitivity test you are testing how sensitive the output is to your assumption of 3% (eg does 3.5% provide a very different output than 3%, or does it make very little difference at all). The assumptions that are sensitive are the ones that should be focussed on with more detail at the calibration stage. But as the notes say if you expect the model to be very sensitive to a particular assumption but it turns out not to be then it may be a flaw in the model which needs to be addressed.

Scenario tests are conceivable scenarios in the real world. Holding all parameters equal except for one, which is stressed, is not a realistic scenario, as many variables are correlated.

I often think of stress tests and sensitivity tests as being useful during the calibration stage, while scenario tests are more applicable to the actual capital setting stage. There is of course overlap between them (eg if a scenario test doesn't give the output that is expected then this may be a flaw in the model), but that might be a way for you to consider them.

Don't forget that when it comes to someone sitting down and building a model, they can't simply build it, apply lots of scenarios and measure the effect. They have to check their work, observe all the nuances of the model based on their calibration, explain reasons for certain outputs, quantify the effect of particular scenarios, etc. Sensitivity tests, stress tests and scenario tests are three means of doing that.
 
Could someone please explain what is the difference between sensitivity testing and stress testing? I'm not satisfied with above explanation fully. Is their application that differentiate them?

As far as i believe sensitivity testing is used to identify those variables which gives large change in output which just small change in assumptions, whereas stress testing is used under loss fitting curve calculation and help us to find the required solvency capital requirement of business and further risk margin by multiplying cost of capital and discounting back using market consistent rate.

Please clarify
 
You must log in or register to reply here.
Back
Top