[Skip to content]

Sign up for our daily newsletter
The Actuary The magazine of the Institute & Faculty of Actuaries

SIAS debates Solvency II internal models

The SIAS sessional meeting on 17 May discussed the challenges faced by GI actuaries in meeting the Solvency II requirements under the internal model approval process (IMAP). Though many of the regulatory requirements for model approval are now in the public domain, it is not yet clear how these should be implemented by firms in practice in many areas.
The debate was chaired by Vishal Desai of the FSA and included a panel of speakers from the GIRO working party consisting of Melinda Strudwick (PricewaterhouseCoopers), Don Johnstone (KPMG), Gabriela Chavez-Lopez (EQECAT) and Stephen Robertson Dunn (Contractor). 

The research for this year has concentrated on model validation, use of third party catastrophe models and the embedding of models in the business.  This update was based on the results of the e-survey (40 respondents across London and Company Market), face-to-face interviews with insurers and views of Working Party members. 

The following are highlights from the discussion: 

Model validation
The e-survey highlighted that model validation (including the production of the validation report) is a key area of activity for firms, requiring extensive additional work and resourcing.  The survey suggested that validation will be completed by a combination of internal work, external consultants and external audit.  Most firms were looking to retain ownership of validation internally. The working party put forward a number of different operating models regarding roles and responsibilities are being used by firms, depending on the scale and resources available. 

The debate then revolved around the difficulties inherent in validating the ‘1 in 200 year' capital requirement, given limited data to support this analysis and heavy reliance on expert judgement. Raza Ullah (Hardy) asked for views on the benefits of producing maps of model processes, data flows and validation and control activities as a starting point for validation of the capital model. 

Karen Seidel (Lloyd's of London) highlighted the key role of independent expert review, given the complexity of capital modelling and the reliance on key expert judgements in the model.   Melinda noted that whilst independent expert challenge was critical to the validation report, in her view there were significant benefits in documenting the model process: both in understanding the model and also in identifying the new forms of management information needed for management to fulfil their validation responsibilities under Solvency II. 

External models
Ms Chavez-Lopez highlighted that an unexpected number of respondents have not made significant progress in their preparation for IMAP in relation to third-party Catastrophe Models.  In particular, the e-survey highlighted that firms saw the following areas as particularly challenging: validation of the model outputs, quality of exposure data, and being able to demonstrate a detailed understanding of a ‘third party' vendor model.  She noted the high dependency on the catastrophe modelling team, and the general feeling of dealing with a "black box", may have lead to a lack of detailed understanding by senior management of the models being used. 

Other Areas

Calibration -
At the current time, the e-survey indicates that ‘Actuary in a Box' is the most popular method employed to measure non-life liabilities over a 1 year time horizon, although there is was representation for Proportionate Emergence.  Stephen Robertson-Dunn noted that the empirical testing done by Martin Cairns (Towers Watson), suggested that the methods produced broadly similar results, when back-tested against historic FSA return data. 

[Note the work that was done last year was to apply each method in turn to some data where we knew what the next result would be (achieved by actually taking the most recent years information out of the data before passing it to the method). The data used was from FSA returns and covered a number of classes and firms, and was only censored for some obviously non-standard issues (such as major calendar shifts, possibly indicating purchase of portfolios etc.). In total approximately 200 data sets were tested.We looked at the number of companies whose actual next year development exceeded the expected amount from the method at various confidence levels. What was found was that the methods showed no strong evidence that any of them were more predictive of each other.]  

Against this backdrop, the Working Party believes that the majority of companies will seek to use the method that is simplest to implement and easiest to communicate to management. 

Documentation -  The working party has reported that firms are expecting on average to substantially re-write 80% of their documentation, with two-thirds indicating that there would be "lots" of new content to be included in each area identified.

As evidenced above, firms are continuing to undertake a significant amount of work for Solvency II. The working party hopes that through the survey's results a look into the current practices being employed might provide some guidance to other actuaries working in the Solvency II arena. 

Please contact Melinda Strudwick or Stephen Robertson-Dunn if you would like to support the working party's research. 

The presentation is available here