Friday, May 17, 2013

(one) Infinite Agent of Change

The use of catastrophe modeling permits participants in the ratemaking process to battle different perspectives of risk through scientific information, each with a with a guise of accuracy.  Competing perspectives view each other as foolish or manipulative while carrying the assumption that a true or real measure of risk is not only possible, but they have succeeded in its identification.

The reality is that different assumptions lead to different measures of risk; and with today's increasing complexity in modeling, risk estimates have no bounds.  A measure of true, correct, or right risk is not possible.

It seems that economic markets (where risk is a good that carries a handsome price) favor the limitless production of risk estimates.  However, public pocketbooks have constraints far narrower than the global risk transfer system.

In my previous post, I argued that Citizens is a mechanism to control the bounds of Florida's hurricane risk.  Where it creates these bounds implies different information and assumptions about future loss behavior.  In doing so, Citizens limits the amount of market contrived risk imposed upon the public as hurricane risk.  Without arguing that Citizens is the best policy tool for accomplishing this task, the alternative is a boundless, infinite conception of risk based on economic market goals.

A recent report from RMS,  describes an upcoming model platform that provides everyone the ability to be their own unique agent of change and demonstrates the "To infinity and Beyond" practice of estimating risk.  The report tells of standard practices in the industry of pricing, selling and trading risk and emphatically highlights that no two entities have the same quantification of risk.  The upcoming modeling platform, known as RMS(one) to be released in 2014, promises to permit for an even more diverse array of assumptions and views of risk.  Here is the executive summary.
Many RMS clients take steps outside of modeling applications to develop their own view of risk— running sensitivity tests, adjusting model loss, and blending model output. Until RMS(one), catastrophe models have not offered insurers and reinsurers the ability to develop their own view of risk, tailored to their guidelines and representative of all exposures. 
The ability to flexibly investigate modeling assumptions, improve the understanding of uncertainty and portfolio sensitivities, and combine losses from numerous sources leads to better business decisions, more efficient use of capital, and ultimately a more resilient risk management strategy. It is no longer enough to have a “one size fits all” approach to risk modeling—our clients need to be confident that their view of risk reflects the uniqueness of their portfolios. Finally, insurance regulators and supervisors increasingly require companies to demonstrate that they fully 
understand the modeling basis for their submissions.  
The RMS(one)™ environment supports (re)insurers in developing their own view of risk, by delivering “open modeling” through the RMS MetaModeler. Clients can incorporate their own research and experience into their modeling; manage model change; prepare market submissions; and meet their regulatory obligations in an efficient, transparent, and documented way.  
Open modeling increases model understanding, enabling our clients to customize RMS models by adjusting or overriding model components and data. Clients can host their own models in the RMS(one) environment—as well as non-RMS models and model results—and can blend model output.  
Through this suite of open modeling capabilities, RMS(one) enables clients to make defensible, transparent, auditable changes to their models and model output in one efficient experience. By gaining control over model assumptions, clients can develop a tailored, complete view of risk at all organizational levels. For the first time, they can confidently validate model applicability on a single platform and can test and modify model assumptions, data, and components, bringing resiliency to their risk management strategy.


  1. Very interesting couple of posts. But.. why should risk be priced in a vacuum, outside the influence of economic markets? If 'risk is risk is risk', then one unit of risk is the ultimate commodity and a market for it represents a very efficient pricing mechanism.

    With respect to estimating risk, the players in the catastrophe risk market (whether traditional insurers/re-insurers, or capital market investors) typically utilize commercial risk models overlaid with their own judgement, data and assumptions - and that is a good thing! To understand catastrophe risk, one needs to take apart those models, understand them, if deemed necessary apply one's own educated judgment, and put them back together again. The last thing anybody needs is for a single, homogeneous, view of risk taken at face value.

    Modeling catastrophe risk is an inexact yet continuously evolving science based on limited data and I don't know of any market participant who truly believes to have the 'correct' number.

    In the long run, all stakeholders will benefit if we strive for more data & risk model openness and transparency, and ever more kicking of the (model) tires.

    1. It is not possible to price risk in a vacuum. The abundance of information available requires decision making about what information to include. But this also means that a public conception of risk is no more right or wrong then anyone else's. Increasing openness and transparency in the modeling process should help to reveal the active decision making process in risk choice, but that transparency should be extended to the public. Right now it is transparent (to a degree) only to those that have purchased a license. As for more data, I don't know that it really matters.

      Public policy about rates is not about the modeled risk, its about selecting a risk that meets publicly held goals. These goals may be different from other economic interests.

  2. I totally agree with you on the need for real transparency. Commercial models are far from transparent, and not completely transparent even for those who license them. The solution would be the development of open-source platforms and data. Now that would require real agents of change :-)


Note: Only a member of this blog may post a comment.