August 20, 2014 20:19
A recent article in August’s Best’s Review, The Rise of the Super Models, by Kate Smith (not Kate Upton, sorry), caught our eye. A lot is going on in the world of computer catastrophe modeling. First, demand by insurers and reinsurers is up and modeling firms are “broadening the scope of risks and regions that they model, with RMS, AIR Worldwide and CoreLogic EQECAT all set to release new models this year.” Among other things, all of the top 3 modeling firms are releasing U.S. inland flood models. This blog has been hard on FEMA and the Corps of Engineers, criticizing the backward-looking nature of flood plain mapping. It looks like the tools to remedy that deficiency will soon be at hand.
Second, modeling firms are shifting from open models to open platforms, which “offer more choice by providing access to models created by third-party suppliers.” According to Ms. Smith, a catalyst for this change is Oasis Loss Modeling Framework, Ltd., an insurance industry-founded and -funded organization. According to Oasis’s webpage, “Barriers to entry have restricted the ability of the insurance community to exploit large elements of available research in hazards and vulnerability.” Such barriers include costs and knowledgeable personnel. The goal then was to create an “open marketplace for models and data leading to much wider access to understandable tools for catastrophe risk assessment.” Open platforms (think Linux) can have great benefits; nevertheless, some are skeptical of Oasis’s practicality in that it is not available for off-the-shelf use, is optimized for an expensive IBM platform, and runs slowly on other platforms, among other things.
The implications for policyholders of all this modeling are three-fold: first, rates; second, policyholders’ own business decisions; and third, others' views of those business decisions.
As insurers better understand the risks associated with particular locations their rates will be adjusted accordingly. This can be a good thing if insurers determine they have overestimated the risk, or if other insurers jump into that market and drive prices lower. But it will be a bad thing if the risk was underestimated and prices rise, or insurers flee a particular market as has regularly happened in Florida and other states. Indeed, at least one state has seen high court approval of the use of models to limit insurance offerings in high risk areas.
Modeling can also be a boon to business planning. What does the future likely hold for a particular location? Will water supplies hold up? Is the flood map reliable or is it outdated? Is the company compounding its exposure by yet another franchise or mall development in a particular region? There is no reason that modeling expertise need be restricted to insurance and reinsurance companies. Other businesses can benefit. However, as pointed out in Super Models, “Models are not a perfect science; there are subjective opinions involved.” Accordingly, businesses should be cautious.
And what if a business does not bring modeling into its business planning? It is likely that if things go awry and the unpleasantness is substantial and can be attributed to an inadequate forecast, an injured party will assert the failure to model the future was negligent. A case in point is In re PXRE Group, Ltd., Sec. Litig., 600 F. Supp. 2d 510 (S.D.N.Y. 2009), aff’d, 357 Fed. Appx. 393 (2d Cir. 2009), where a reinsurance company found its failure to rely on a particular model was the gravamen of a class action plaintiff’s security fraud suit. PXRE was a thriving reinsurance company, whose business was conditioned on maintaining an A- rating. Unfortunately, Hurricanes Katrina, and then Rita, and then Wilma, devastated certain portions of the Gulf Coast to its reinsureds’ detriment. PXRE stepped in and paid on its reinsurance contracts but the losses kept increasing. It relied on models to reassure the investment community that it remained financially sound in order to raise money. The models it relied on, however, turned out to be inaccurate, and ultimately PXRE's rating crumbled and it succumbed to the unprecedented losses. The class action ensued.
Plaintiff claimed, among other things, that PXRE should have relied on a higher estimated loss ($40-60 billion by RMS) rather than valuations of $30-40 billion touted by PXRE’s own models as well as by ISO and Air Worldwide. The district court opinion gives a lengthy dissertation on the standards to be applied in a securities fraud case on a motion to dismiss and concluded that PXRE was not reckless in its reliance. More germane to the issue here, is that PXRE was able to defend itself because it had relied on models.
Granted, modeling was part of PXRE’s business and, no doubt, a lack of modeling would have been reckless. But, is a prudent non-insurance business going to eschew modeling on the theory that no one else in its industry relies on them. If models are becoming more widely available, as suggested by Super Models, the path of the prudent business is, at the very least, to consider whether modeling has something to offer.