C. Warren Axelrod

The Personalization of Risk

I realized when I received several comments regarding my September 12, 2011 column “Risk Mismanagement – Scoring vs. Monte Carlo vs. Scoring” from Doug Hubbard and others, that I hadn’t been clear enough in my description of what I had termed “subjective risk.” It also seems that it was not readily apparent to readers whether I supported risk scoring or Monte Carlo methods.   So, let me try to clear up these misunderstandings.

First, the easier one … I am a strong proponent of  the Monte Carlo approach to risk assessment. For a multitude of reasons, many of which are listed in Hubbard’s book, scoring methods are deficient as a means  of expressing risk. Risk scores are highly subjective, not readily able to be aggregated, carry different weights in the minds of various assessors, etc., etc. Yet for  all their deficiencies, risk scoring remains hugely popular. And that’s because risk scores are easy to come up with and simple to present. They don’t involve  complicated probability theory that is difficult to understand and harder to implement. While I believe that scoring, if used with full knowledge of its  limitations, can have some value in focusing management on particular areas of risk and can be used to indicate some broad measure of relative importance, risk  scoring does not properly represent the nature of risks with respect to value, uncertainly and time.

By the way, I must also correct my characterization,  in my September 12, 2011 column, of the OCTAVE® approach being “totally dependent on scoring.” In fact, various OCTAVE® publications play down the use  of scoring, favoring the use of high-medium-low categories—this is what I have used in many of my own publications. To be fair, the OCTAVE® researchers do  introduce scoring in a few places, but with many disclaimers and warnings against its ubiquitous use.

Post a Comment

Your email is never published nor shared. Required fields are marked *