Risk modelling - Underwriter responsibility: Model behaviour

In the long history of underwriting, risk models are a relatively new development. Nevertheless, they have quickly become indispensable tools for underwriters in the assessment of risk appetite and capital setting as well as in the appraisal of premium levels. In many ways, models have proved an extremely positive force in the insurance market, enabling underwriters to assess risk with increasing accuracy and in a more standardised way.

Solvency II is already challenging the industry and calling into question the current application of models. But fears that the new capital adequacy regime will force underwriters to abandon their trusted systems in favour of a centrally determined model are overrated. Solvency II is not intended to make underwriters entirely dependent on models — or even to force underwriters to subscribe to a particular form of model. It is intended to standardise the capital-setting process and increase the underwriter's responsibility in the way they approach modelling.

The introduction of advanced modelling techniques has allowed data metrics to be streamlined across the industry, approaches to risk to be formalised between players, and data requirements for projecting risk to be standardised. These developments have allowed for a greater number of complex risk scenarios and eventualities to be projected and incorporated into capital allocations. And, although they are not crystal balls, when combined with an underwriter's skills and experience, the information gleaned from models can provide an edge that allows insurers to diversify their risk.

Simplified world view
However, while models are excellent supporting tools they are merely mathematical representations of real-life scenarios, and each model therefore only offers one, simplified view of the world. Yet one of the biggest challenges for insurers and modelling companies alike is accounting for, in Donald Rumsfeld's words, the 'unknown unknowns' — potential risks that are not being covered in the model. For example, to neutralise the 'unknowns', Ascot decided to develop its own in-house systems that work in tandem with the models to provide a safety net of 100% aggregates in key zones.

As clients expand their range of activities and insured values, they are demanding that insurers keep up with their growth and associated risks. The innovation of new product lines and underlying terms puts additional pressures upon our models and modellers to ensure they remain valid.

This is, of course, a greater challenge in some areas than others — for example, the growth of the terrorism product in the past decade has led to large portfolios within wholesale insurance and reinsurance, which are exposed to a peril with little empirical data upon which to base a model. Similarly, the stress tests we require of our models must also pass muster for more obscure risks such as those assets sited offshore.

The past few years have been a challenging time for the industry's assessment of offshore exposure. While grappling with a changing catastrophe landscape — which has seen hurricanes Ivan, Katrina, Rita and Ike providing a wealth of new stress tests for insurers — the industry has also had to absorb the impact of both technological innovation and the demand for fresh hydrocarbons, pushing our clients into increasingly deeper water and more testing operating environments.

Aside from wrestling with the nuances of offshore vulnerability, models must also ensure they are complete — capturing all relevant exposures. The recent losses of West Atlas and Deepwater Horizon represent textbook cases whereby the offshore property damage represents just a small component of the overall loss. How these are represented in our view of risk is of significant importance.

Given the potentially larger role for models once Solvency II comes into effect, it is likely that insurers will up the pressure on modelling companies to provide both increasingly greater scope as well as more accurate and timely representations. Although it is unrealistic to expect our models to be infallible, it is not unreasonable for insurers to expect that the model outputs are both reliable and stable.

The widespread use of models has given the risk modelling companies a significant amount of power to influence how business is written. So, while Solvency II encourages increased model use, it also demands that insurers understand what is covered by the models and what is not. Senior management, including a company's board, will need to be aware of the inherent risks involved with a given model and the steps being taken to compensate for any gaps.

The process of preparing for this European-driven regulation by the insurance industry has been daunting and spawned an industry of its own, dedicated to helping companies meet the requirements. Discussions centred on the role of models in Solvency II have highlighted a perception that the new regime's attempts to standardise risk assessment will boost the authority of models and limit an insurer's ability to differentiate their business.

But, at its heart, Solvency II is not asking insurers to do any more than what should already be happening: ensuring their internal systems and models are linked up and that the view of risk is both translated and supported in their capital projections. The objective is standardisation in rationale, not necessarily in your business or risk appetite.

Working in tandem
Models work best in tandem with an underwriter's skills and experience of their particular class of business. Within the regulations there is some flexibility allowing insurers to demonstrate the effectiveness of their systems and accuracy for assessing capital requirements. If an insurer can prove that its existing systems are joined up and working together, logically and profitably, Solvency II allows for the possibility of continuing with business as usual.

Models are not intended to act as failsafe predictors of risk; they are tools to be relied on but not depended upon. As underwriters, our role must remain to set pricing and premium levels on the basis of a thorough consideration of risk and not solely on the output of a model. Solvency II could be a turning point for our industry and an important part of the change will be to increase the responsibility on underwriters to better understand risk.

The incoming regime will allow us to embed capital models, which if done well should best represent our business. Models may have attracted their fair share of bad press over the past five years, and the industry should of course be cautious of their shortcomings, but it should also be equally cynical of the alternative — to set capital requirements with crude damage factors by line of business or country would be regression to the middle ages.

Matthew Washington is head of aggregation & reinsurance at Ascot Underwriting

Source: Post - 20 January 2011