PRA on General Insurer Technical Provisions under Solvency II - Taking the "TPs"?

Allow me to take a quantum leap outside of my comfort zone while I pick my way through the PRA's latest Insurance Industry aide memoire, via a consultation paper on the calculation of technical provisions in General Insurers.

This looks specifically at TP calculations with Solvency II in mind, and is aimed specifically at GI firms currently in IMAP. That said, the tone and technical matter covered is an excellent heads-up to Actuarial, Risk and model validation personnel currently active in this space about how the PRA approach to assessing Solvency II compliance is developing.

The document itself reads very much like their last consultation paper release on Deferred Tax Assets, insofar as it is a laundry list of "what not to do" - look at how many times the expression "should not" appears! They have leaned on their findings from both thematic reviews of TP calculations (Life and GI-specific Questionnaires were sent out a year ago) as well as from IMAP and ICAS, so their finding will be well supported by most recent practices in the UK.

The consultation window is pretty short as well, with a mid-April shut-down scheduled, so if you don't like the cut of their jib, you'd better speak soon.

Stand-out points for me included;

Generic


ENID - TP accommodation required

• Expectations of Delegated Acts content are cited throughout, but in terms of the exact date of their public provision, they can only go with "Q3 2014". From what I have seen, there is nothing cited which isn't in the November 2011 draft.
• The abandonment of the term "binary events", replacing it with "Events not in data" or "ENID" - the fait accompli of "binary" (that events which are not in a data set must therefore be extreme and/or rare) is confirmed as unacceptable.  The PRA don't appear to be wedded to the old term in any case, and while the actuarial profession used it liberally in the past (here and here for example), they began a transition away from it late last year (p45 of this).
• "Any data that can have an impact on the outputs of the internal model should be considered to be 'used for the internal model'" (3.19) - important IMAP message across sectors I think!
• There is evidently some concern that firms are thinking of relying on the work of external model providers to meet Solvency II standards, with the PRA confirming that firms may not rely on "...generic validation performed by the model vendor" (3.25). This means that the model validation relationship between IMAP candidates and their third-party providers needs to be much more invasive and aggressive, and needs to start pretty soon!

Technical Provision-Specific

• A large number of points made in the paper relate to over-simplifications, which should help anyone who is struggling with the concepts of materiality and proportionality. These include methods relating to ENID, Risk Margin calculations, Approximations and  the emergence of risk over one year
• Similarly a few tricks of the trade appear to have been scuppered, such as using optimistic business plans for setting provisions, "actuary in a box" methods and assuming improved underwriting performance
• Some substantial focus around the quality and quantity of challenge applied to External Models (focused on third party Catastrophe models in this instance), in particular the challenge of  assumptions used by the provider (3.16-17 and 3.26-28)
• The concept of "cumulative materiality" is introduced in the context of multiple approximations, a concept which I suspect many firms are still struggling with in the context of Internal Model change (2.9)
• An interesting take on the justification of assumptions, with the PRA taking umbridge with firms using "industry standard" or "established good practice" as a supporting argument, rather than using their own risk profile as the basis for support (3.15)
• A section which seems to advocate conservatism, if not prudency, in the setting of sensitive parameters (3.10), as well as advocating the use of stress and scenario testing to make up for ENID when setting parameters (3.2)

Certainly lessons for both Life and GI internal model applicants in here, and the PRA should be congratulated for getting this paper out in good time. I'm not necessarily convinced though that third-party providers of internal model inputs will happily acquiesce with the demands which the industry are being asked to make of them here.