Industry Good Practice for Catastophe Modeling & Solvency II – A Perfect Opportunity for Review: Part I, Background and General Principles
The UK insurance industry, supported by the Association of British Insurers (ABI), has developed a report “Industry Good Practice for Catastrophe Modeling & Solvency II” to guide companies’ use of catastrophe modeling under Solvency II. The guide consists of three main sections:
- General Principles: governance and documentation - the suggestion being that it is management’s responsibility to ensure robust systems, controls and documentation are in place to understand and gain comfort in making decisions using model output. Outsourcing elements of the process should be governed by policy and a specific outsourcing service level agreement.
- Operational Principles: data and processes - data should be tested for accuracy, completeness and appropriateness, and data manipulation should be documented in a formal data policy. Model validation and selection should be justified and clearly documented. Sensitivity testing should be important in the application of ‘modifiers.’
- Technical considerations: model approach and uncertainty - the report suggests companies develop a “bespoke view of risk,” whether based on a single model or a blended multi-model approach. A company should show awareness of uncertainties within the model and demonstrate that decision makers understand the implication of uncertainties.
Guy Carpenter is ready to help with education and training, catastrophe modeling documentation (including extraction of documentation from model vendors) and data assessment. Key amongst our services will be our expertise in providing model suitability assessments (MSAs) with which we can guide clients in selecting and using catastrophe models.
There has been considerable discussion for sometime about how Solvency II will impact the way catastrophe models are used in internal models (whether partially or fully implemented). Ever since Solvency II principles were first published, (re)insurance technicians have been reading the relevant articles diligently and trying to translate guidance into how existing practices should change. Through our analysis and during discussions with our clients, we have learned that this is not a simple exercise. By their very nature, catastrophe models aim to quantify the impact of severe and infrequent events, making them challenging to validate. This same challenge makes it difficult for senior management to confidently use the results from the models in their risk management processes.
For some time there has been a general consensus that Solvency II would put more emphasis on understanding catastrophe models (including their limitations), the systems and controls around the whole process and, of course, the documentation of each step along the way. Indeed, the UK Financial Services Authority (FSA), through the Individual Capital Assessment Standards (ICAS) regime, began to focus on the reasons why a particular model would be appropriate for a particular insurance company, in addition to the model output. The eagerly awaited publication of the ABI-supported report on ‘Industry Good Practice for Catastrophe Modeling & Solvency II’ provides an important opportunity for (re)insurers and their service providers to take stock of current practices with a view to being prepared to face implementation when Solvency II arrives.
The report is the culmination of the work of a wide group of contributors including insurers, reinsurers, vendor modeling companies and reinsurance brokers (including Guy Carpenter). The quality and collaboration of the contributors - and the comprehensive coverage of the relevant issues - make this report essential reading for those concerned with catastrophe modeling. Here, we provide a quick summary of the report for you, along with a brief discussion of possible implications and finally insights into how Guy Carpenter can support your efforts in gaining confidence in the results and complying with Solvency II.
Section 1 - General Principles
The first section of the report focuses on the governance and documentation principles that companies should demonstrate when using catastrophe models within their internal model frameworks.
The report points out that any organizational changes should be proportionate and appropriate to the company but advocates ownership of catastrophe model utilization at the board level. It also indicates a structure that includes a chief risk officer (CRO) who is backed by catastrophe risk specialists responsible for the quantification and communication of catastrophe exposure throughout the organization. This should not pose many problems for companies preparing for Solvency II, as most have been moving toward a version of this structure in anticipation of the Pillar 2 risk management requirements.
It is also prudent for senior management to have an overall understanding of the organization’s catastrophe exposure and how it is modeled - through transparent reports and presentations. The process should convey a broad understanding of the strengths and weaknesses of the models used, according to the report. Further, it suggests that it is management’s responsibility to seek out information on the models to a level of detail that enables them to feel comfortable making decisions based on their output - in light of the level of uncertainty inherent in the results.
Getting to this point is likely to involve a significant education program for many companies. In-depth catastrophe model knowledge has long been a bastion of the few technical individuals affiliated with reinsurance, risk management or capital modeling. Transferring this knowledge up to senior management, in appropriate detail and at regular intervals, should become business as usual.
The report highlights that the process of incorporating catastrophe risk into an internal model doesn’t start and end with running catastrophe models. Rather, robust systems and controls should be in place from data sourcing, cleaning, modeling, validation, interpretation and inputting results into the internal model - right through to decision making. These systems and controls should be documented, audited and followed. How this workflow changes when new model versions are released should also be carefully considered in advance and built into the planning process.
One of the areas that will be read with interest by most companies is the section on “The use of third party service providers.” Many companies do not have the resources or the expertise to license and run vendor models in-house and instead rely on outsourcing much of their catastrophe modeling to reinsurance brokers. By outsourcing this activity, however, companies cannot outsource responsibility. Ownership of the results and how they are used should rest with company management. The dilemma presented, therefore, is how to oversee the entire process adequately and gain the level of insight necessary to make robust decisions in light of all the information while not engaging directly in the activity itself. The solution proposed in the report is to have in place both an outsourcing policy and a specific outsourcing agreement. The former should address why, how and to whom a service is to be outsourced and include considerations of any potential conflicts of interest and clear descriptions of how roles and responsibilities are divided between the service provider and the company. The obligations of the service provider in general are currently covered in existing broker service level agreements (SLAs), but it is likely that a dedicated legal document will be needed to encompass a more detailed specification of catastrophe modeling services, and in particular the responsibilities of the company receiving the services.
Anyone searching for respite from the seemingly endless ability of Solvency II to generate documentation will, alas, not get any relief from reading the report. It does, however, offer some very sensible advice on indexing, version control and perhaps most importantly, structuring the material in a way that makes it easy to mine for readers who require different levels of detail. There is also some valuable guidance on content that will make collating existing documentation and identifying any gaps much easier.
The provision of documentation (or, potentially, the lack of it) from model vendors has been an area of contention recently, and it is still not resolved. Documentation is subject to restricted distribution due to the proprietary nature of the material, and Solvency II places no obligation on the vendor modeling companies to supply their documentation to non-license holders (or the regulator, for that matter). This creates an issue for non-license holders who access the vendor models through a third party. The report encourages model vendors to become more transparent. Up to now, though, the response to this topic from some of the vendors has been less than encouraging. The fact remains that it is the responsibility of the company to satisfy itself regarding the validity of a model. If the information is not available, the company may need to move away from that model.