Natural disaster management: model uncertainty

Normal view (turn off text only mode)
You are here:

NATHAN Risk Suite - Munich Re

Model uncertainty – Are you building on the relevant scenarios?

Despite improvements in the scientific understanding of natural perils in Australia and New Zealand and our ability to capture and analyse ever-increasing amounts of data, in catastrophe modelling uncertainty remains a fundamental notion – with a strong impact on risk and capital decisions.

The role of stochastic catastrophe models in the (re)insurance industry has evolved since the release of the first commercially available models in the late 1980s. In fact, recent events and new scientific advances have shown us that the degree of uncertainty in the results produced by catastrophe models is even bigger than originally thought. This has led to a growing focus amongst regulators and rating agencies on how companies address the issue of model uncertainty when making capital and catastrophe risk decisions.

“APRA wants each insurer to challenge itself about its governance and management of catastrophe reinsurance arrangements and to adopt very good practice. We particularly want the insurer to understand the strengths and weaknesses of any models it uses, and the degree of uncertainty in the results produced. And we want model outputs to be complemented by significant further analysis. Lastly we want the insurer to satisfy itself that the residual risk is truly within its appetite.”

Ian Laughlin - Deputy Chairman Australian Prudential regulation Authority

Uncertainty arises from all components of a catastrophe model. The hazard calculation, the vulnerability assignment and the representation of the exposed values contribute to the overall model uncertainty. Even where a long historic record exists, its accuracy is often questionable. In addition, the time span of such data is way too short to cover the full range of realistic scenarios. The Tohoku earthquake, for example, came as a complete surprise to the modelling companies despite the very long historic record in Japan.

Beware of wrong accuracy

In many countries, vulnerability functions are mainly based on engineering calculations and are not constrained by loss data. The tendency of using ever-more detailed vulnerability functions for specific building characteristics is aggravating this issue and creates a sense of wrong accuracy. The assumption in the exposure data input to comply with all these parameters has a major impact on the modelling results. After catastrophes it becomes clear that wordings or valuation aspects were not adequately addressed.

Good corporate governance demands that companies explicitly consider this uncertainty in model outputs, with risk and capital decisions reflecting this understanding. Greater focus is being given to data quality, reasonableness checks against recent events and the setting of catastrophe reinsurance limits that make allowance for the uncertainty in the modelled outputs on which they are based.

Munich Re´s modelling approach

When addressing model uncertainty issues, prudent exposure management and understanding of risk is essential. This is particularly the case for one of the largest exposures an insurance company faces: natural catastrophe risks. At Munich Re we have developed our own proprietary internal natural catastrophe models that provide us with an alternative view of risk.

In addition to modelling capabilities, Munich Re provides expertise, global experience and a multitude of tailored reinsurance solutions for frequency and severity exposures. Reinsurance can play a vital role in an insurers’ capital management plan through capital enhancement, substitution or efficiency.

Risk factors

Climate change, natural climate variability, natcat and resilience – find out about the relevant risk factors.

Related topics

Download

Natural hazards in Australia and New Zealand

NATHAN Risk Suite - Munich Re

Floods, cyclones, hailstorms, bushfires, earthquakes and volcanic eruptions – get a detailed overview of the science of natural hazards and their financial impacts.

Understanding risk

NATHAN Risk Suite - Munich Re

Risk is the product of (the probability of) a hazard and its adverse consequences. See how it affects your business.

Contact


Main Navigation
Service Men
Accessibility

Note


This publication is available exclusively to Munich Re clients. Please contact your Client Manager.