Solvency II –the PRA does want to approve internal models


Insurers are on track to implement Solvency II (SII) by 2016. But getting approval from the regulator for an internal model is still proving a daunting challenge. An internal model – the processes, systems and calculations that quantify and rank the risks faced by the business – doesn’t have to get approval. But for those who choose to go down this route, it provides essential groundwork for implementing SII.                                                                 

The PRA may naturally have some anxiety over approving internal models – giving firms the ability to set their own capital is no small thing.  There is a strong sense however, despite what some say, that the PRA really does want to approve internal models and help firms get to where they need to be. 

Some important questions insurers are still grappling with include:

How much documentation do I need to support my assumptions? My view is that the amount of documentation is immaterial. The important thing is to articulate your thought process and provide evidence that your approach is better than anything else.

What supporting documentation do I provide around dependencies? When it comes to dependencies, the PRA expects firms to do better – both in terms of the justification of parameters but also the understanding and methodology comparison for tail dependency i.e. that when things go really wrong, they tend to go wrong together.

Should we focus more on justifying or validating our internal models? The PRA make a clear differentiation between justification (for model inputs, parameters, assumptions) and validation (for model outputs and alignment with the risk profile of the firm). Some have questioned whether the PRA is really just focussed on the model output “number”. I don’t think that is accurate – either way, the emphasis on meeting (and evidencing of) the Solvency II Tests and Standards, and the focus on justification are coming through much more strongly – reinforcing a number of messages we discussed in this paper.

Prudence and conservatism – could it reduce the justification burden?  I would advise against it for significant areas such as risk emergence because the model is being approved for future use. Appropriate methods are really needed, particularly because the business could change and alter the risk profile of the firm.

While much has been done to progress internal models, there is also an awful lot more to do. Perhaps agreeing the most difficult and contentious areas with the regulator naturally remain outstanding – it’s going to be an interesting ride.


Read more articles on