The degree of uncertainty for modeled perils draws a lot of attention from most insurance companies. A significant amount of resources are put into improving data quality, modeling assumptions, model selections and other related items.
Executive Summary"Don't let great get in the way of good," writes Hanover Insurance Group's Prateek Chhabra, who advises that rudimentary approaches for assessing and quantifying nonmodeled risks are better than none. Here, he presents a simple framework for risk assessment, explaining that carriers can turn up the dial on quantification techniques—increasing levels of sophistication and resource deployment for risks that are material to the company and those that are actionable through pricing, underwriting and exposure management decisions.
For most companies, these efforts are appropriately spent as the modeled perils are usually the largest contributors of underwriting risk. As a result of this focus, however, smaller frequency and severity risks, which may be significant, are often not modeled and therefore are inadequately addressed.
Nonmodeled risks are defined as risks that do not have a traditional structured and systematic way of identification, assessment and quantification to incorporate in the underwriting, pricing or risk management processes.
Lower perceived materiality coupled with the complexity associated with developing independent views, models or methodologies often seems like a daunting task resulting in this inadequate treatment. This leaves potential for gaps in risk assessment and management.