Tuesday, 30 October 2012

Aon Benfield's CRO guide to Solvency II - in case you're not ready yet...

For all those CROs who are about to get left holding the Solvency II baby three years early by their over-enthusiastic executive colleagues, Aon Benfield pulled together a CRO guide to Solvency II which aims to take the journey "from complexity to best practice". 10 out of 10 for ambition...

It leans heavily towards General Insurers/Reinsurers (indeed it reads like a reinsurance sales brochure in many parts!), but nevertheless contains a suite of very useful content for anyone in the Risk space, as well as attempting to shatter a few myths. I took the following from it;
  • Steady early bits on capital planning and common questions a CRO should be posing in that space
  • On page 5, an excellent table comparing standard formula against internal modelling by risk driver, in particular emphasising why internal modelling may be more appropriate, rather than how much capital it could shave off. Being able to explain to the national regulator why one has neglected to apply the enhancements that internal modelling introduces to the accuracy of one's quantitative risk profile would be a smart thing for CROs to practice!
  • The undo some of that noble work by suggesting part of any IM feasibility study should include estimating the capital benefits!
  • Nice examples at the top of p6 of what mixes of business lend themselves to benefitting from an IM approach
  • Highlighting that domicile of firm continues to dictate feasibility of IMs for smaller firms (i.e some countries can't staff it!).
  • Recommend reviewing SF SCR factoring in the draft L2 asap. As was clear from the E&Y research I covered yesterday, many firms across the EU consider themselves to be advanced in the Pillar 1 space while disregarding draft L2. They highlight the Swiss experience as one where they struggled to authorise models for "Day 1" approval, and the Aon crowd propose some meaningful contingencies on p8
  • Useful analysis of capital drivers and optimisation strategies (p9-10)
  • Section on expert judgement validation (p15), touching on the Level 3 expectations, and in particular how a (non-Actuarial) CRO may struggle to adequately challenge certain judgement calls, such as selected data series or correlation matrices, without specialist advice. Very hard for smaller firms to obtain that, as most of their actuarial function will have probably contributed to the judgement!
  • Note that one of the key challenges for documenting the IM is getting the best-placed people (who are normally swimming in BAU) to pick up a pen and write!
  • Neat section on ORSA (p27-29), emphasising that SF firms with complex risk profiles may find they struggle to justify that approach when concluding the assessment. They go on to suggest that early experiences of ORSA Report/process documentation submissions have left CROs feeling that the regulatory approach is (Level 3?) tickbox as to content expectations.
  • Key challenges for CRO in briefing and educating senior colleagues for Solvency II-readiness are all fair, in particular the gap that could emerge if a CRO is not also an executive member.
  • The section on Risk Appetite is particularly useful for smaller non-IMAP firms, who may struggle to quantify their target measures - whether using Standard Deviations/volatility measures as suggested is a touch too simple depends on the business I guess.
  • The Pillar 3 section hits on the same issues I (and the FSA!)have picked up on earlier, such as end-user computing, inability to transition to BAU, data ownership issues etc.
I did take exception to a couple of bits in here, where the industry or indeed common sense appears to suggest otherwise;
  • The "fallacy" outlined on p5 that an IM enables a firm to hold less capital than an SF equivalent. The research I pointed to yesterday (p20) suggests across the EU that modellers are already "making it rain" with their capital savings
  • That the IM alternative for Op Risk is based on ORIC and individual loss event info. I'd certainly seen Milliman suggest that this approach is as flimsy as the SF approach, recommending options such as Bayesian networks to generate IM inputs.
  • Concerns that evidencing senior management model "use" could create a "value-destroying documentation burden". Is that what we call "minutes" these days!
  • Comments around the documentation delivery for the Internal Model Application Process becoming detached from the underlying processes referenced in those docs influencing BAU value-adding activity are perfectly valid, but no real solution is proposed.
  • The operation of the Model Change Policy features heavily (p19-21), as anyone in that space would expect. Again. little offered in the way of solutions, but I certainly would have expected more discussion on the "scope" of the model, which in my experience is a solid, liquid or gas depending on which control function you speak to, and I'm sure the FSA would agree!
PS All the best to you guys on the US East Coast, let's hope the worst has passed...

No comments:

Post a Comment