Monday 18 June 2012

Lloyds of London on Validation - Workshop Output

For all of you working on one of the FSA's (and indeed I suspect other supervisors') pet peeves, Model Validation, the Lloyds guys published some slides from their recent workshop on the topic.

Anyone not in the IMAP space may still find their presentation of interest, particularly as they have already had to ask for 3 months leeway on their original application "landing window", perhaps hinting at the enormous complexity and expense of producing material to support a successful model application.

They start with a "Top 5" ways to improve existing validation practices, rather worryingly stressing coverage and ranking, which one would think would be par for the course rather than something which required additional emphasis. They go on to hint at, if not confirm;
  • That both analysis of test results and subsequent escalation has been found to be lacking during their validation dry run
  • That historical syndicate experience is "always necessary" in modelling, even if not sufficient on its won
  • An over-reliance on sensitivity testing using pre-defined SCR ranges
  • Their recommended materiality hierarchy of validation tests
  • Their take on reserving risk (not my forte, but dig in if it's yours)
  • Their take on correlations and the diversification matrix (interesting bit on diversification in the tail)
  • That their model walkthrough exercises brought up material issues for "most" agents, and that clearly some syndicates have been asked to complete remedial action before the end of this month
  • That they are waiting for L3 before reviewing their validation reporting formally.
Not sure whether this keeps the guys on track for their group application, but certainly not an unhealthy thing to be picking up such flaws at this stage - if it was that easy, everyone would be doing it!

No comments:

Post a Comment