Hello, In April 2020 attempt, answer to Q2 iv) talks about stresses in internal model. My question is- how do we calculate internal model in reality? Standard formula is very clear as unstressed-stressed net assets but I did not think of the same definition for internal model as its definitely not explicitly mentioned in Core Reading. I dont have practical experience on Solvency II. Please help. Thank you!
Hi - if an insurer is calculating its SCR using an internal model rather than the standard formula approach, it is basically using the same underlying method as for the standard formula, ie assessing the impact of each extreme risk event on its basic balance sheet (since the SCR still has to be calibrated to (at least) a 99.5% one-year VaR) and aggregating the results. The main differences are that it can choose its own stress events for each risk, and is not constrained to aggregate using the stated correlation matrices (eg can use a more sophisticated modelling approach that models the stresses together, perhaps using copulas to describe the dependencies). There are other more sophisticated modelling impacts that can be allowed for with the internal model approach, as described in the Core Reading (eg dynamic hedging).
Hello, In addition to choosing own stresses and correlation matrices, does the insurer also have discretion with respect to the risks for which it will hold capital if they choose internal model? Lets say, if insurer has exposure to climate change risk and currently, as SII does not have an explicit requirement to hold climate change risk SCR, can insurer hold additional capital for this risk? Will this be possible even for standard formula? Alternatively, is there a possibility for holding capital for fewer risks if insurer uses internal model? Hypothetically, say if insurer has minimal exposure to corporate bonds hence negligible credit spread risk, is it possible to not hold any capital for this risk under internal model/standard formula per Solvency II?
Also, when analyzing change in SCR, what does 'change in 1 in 200 yr calibration' and 'inclusion of different risk factors (if using internal model)' mean?
Yes for the full internal model - provided it covers the prescribed list of risk groups that the SCR has to cover. And yes if it adopts a partial internal model (part standard formula, part internal model). [Bear in mind that climate change risk impacts are implicit within other risk categories, including insurance and market risks.]
If using an internal model, the company might change its view of a '1 in 200 adverse event' for a particular risk. In which case, the SCR would change once it has been re-run on the amended stress. Similarly if different risk factors are included in the SCR calculation, the SCR will change.
Hi - yes, if an insurer has zero exposure to a particular risk it would have a zero contribution to the SCR from that risk. I guess there's some point at which 'minimal' tends into 'zero', and it may be that some risks can be ignored as being immaterial, but I wouldn't like to say exactly where that cut-off point would be!
