March 7th, 2019

Building for Resilience: How to Avoid a Catastrophe Model Failure: Part IV

Posted at 1:00 AM ET


Imelda Powers, Global Chief Catastrophe Modeler


Valuation Assumptions

There is no database of agreed property, contents or business interruption valuations among vendors or insurers.  Any user’s particular valuation may come from databases of property prices, or rebuild values from claims adjusters and building surveyors. If the model’s assumed valuations are under- or over-estimated, then the damage function may over- or under-compensate in order to balance to historical industry event losses during the model-building process. Consequently, it is important that the user adjust the damage module to reconcile these differences.

Missing Components

In addition to adjusting the “off-the-shelf” model results based on the above exercises, users should also compensate for components that are missing from the model.  Below are some examples:

(1)     Portfolio exposures - Some exposures may not be recognized (for example, infrastructure risks like bridges or tunnels, or outdoor structures like gazebos, carriage houses, etc.) or are misrepresented because of poor location information, such as non-geocoded risks.

(2)     Portfolio changes - Consider adjustments to modeled losses for a growing or shrinking portfolio, or one with shifting geographies or risk types.

(3)     Sub-perils - Examples of these include storm surge in Ireland, France, Belgium, the Netherlands, parts of the United Kingdom and Australia or tsunami risk resulting from earthquakes.

(4)     Policy terms - Terms and conditions covering sub-perils cannot always be adequately addressed in today’s models, so post-analysis adjustments should be considered.

(5)     Policy leakage losses - Storm surge losses are not often covered within a policy, but adjusters may have difficulty separating wind versus surge losses for some properties.

(6)     Coverages - Examples include debris removal, building code upgrade, land improvement and contamination.

(7)     Loss adjustment expenses - Costs to adjust claims may increase as a result of the complex nature of extreme events.

(8)     Post-event loss amplification - The post-event loss may increase because of higher labor and material cost or unusual delays following an extreme event.

(9)     Cascading impact - Indirect loss caused by failed risk mitigation measures, such as the malfunction of cooling generators in the Fukushima Nuclear Power Plant after the Tohoku Earthquake or fires in Breezy Point, New York following Superstorm Sandy, can increase loss.

(10)    Changing regulations - New rules governing how insurers adjust claims can impact loss reimbursement; for example, the application of deductibles in Windstorms Lothar and Martin and payment for slab cases in Hurricane Katrina.


If an insurer has performed a macro level analysis of its model - examining the various components carefully and supplementing missing or inadequate modules appropriately, while also leveraging micro level review to construct a fit-for-purpose, tailored approach - that minimizes the possibility of model failure. Adapting models to unique business opportunities may support profitable growth.

Read Part I >>

Read Part II >>

Read Part III >>

Click here to register to receive email updates >>

AddThis Feed Button
Bookmark and Share

Related Posts