This rebuttal was penned by E. Stuart Powell, Jr., MA, CPCU, CIC, CLU, ChFC, ARM, AMIM, AAI, ARe, vice president of insurance operations with the Independent Insurance Agents of North Carolina Inc in response to Dr. Lawrence Powell's recent series on the importance of credit scoring.
********************************************
Some thirty years ago, I read a book about near death experiences. The author had interviewed a number of people who claimed to have had a near death, out of body experience. He described the common characteristics of the experience. One criticism of the book was that these people were probably dreaming or were affected by some type of suggestion. For me the problem with this criticism was that these people did not know one another, had no contact with each other, and appeared to have nothing in their background that would explain why they each described a common experience. Either each had, in fact, experienced a common phenomenon or they were participating in a common dream. Most people would not find the common dream explanation credible unless they were acquainted with the writings of Carl Jung.
The problem I saw after reading the book was how to determine the credibility of subsequent reports of near death experiences after we all learned what the characteristics are. How could I tell whether my experience of near death was real or a dream emanating from a suggestion the book made to my conscience or sub-conscience? It appeared to me that the book's publication had undermined the method of establishing the credibility of the experience.
Dr. Lawrence Powell's article, "The Impact of Credit-Based Insurance Scoring on the Availability and Affordability of Insurance," is well researched and written. It is a compelling argument for the use and benefits of credit scoring in insurance underwriting and pricing. (I am so impressed with his scholarship and intellectual acumen that I am convinced he is a distant relation.) Dr. Powell, however, does point out one of the fundamental concerns that non-statistical types have with credit scoring. Namely, the relationship between the credit score and insurance losses is a correlation without cause and effect. In other words, it might be possible to determine that blue-eyed people live on average three years longer that brown-eyed people. The color of the eyes, however, is not the cause of the difference in mortality. It is coincidental. In the insurance sense, the good credit score may not be related to good driving habits or above average home maintenance. There is a correlation between a credit score and insurance losses but one does not cause or result from the other.
Could this characteristic of credit scoring cause an external factor to effectively modify the correlation and how would we know. Now that people know the credit scores are so important, will efforts to improve their credit score (the media are ripe with vendors offering this service) cause a disruption in the correlation? And if so, how will the insurance underwriter become aware of this disruption? Does the knowledge that credit scores affect insurance cost cause behavioral changes that will improve the credit score without changing the underlying insurance risk characteristics?
It would seem to me that to the use of a statistical correlation that has no cause and effect relationship between its components is a potentially dangerous practice. A change in the correlation precipitated by the knowledge of the correlation could cause an undetectable change in the benefit derived from the correlation. (I cannot believe an insurance agent wrote the previous sentence.) If I can improve my credit score by conscious effort without changing my driving habits or home maintenance habits, I will be getting an insurance premium that is inadequate for the expected losses resulting from my unchanged personal habits.
I realize that most of the people running today's insurance industry grew up in an era where the credo was, "If it feels good, do it." Most of us by now have realized that it may be a good credo for a few years during late adolescence but it is not necessarily a credo for a long and health life. Maybe I am all wrong in my concern, but the long-term viability of a statistical correlation that represents no true relationship with its constituent parts seems problematic.
**************************************
The views expressed in this article/commentary are solely those of the author and do not necessarily represent the views of MyNewMarkets.com, the Insurance Journal or Wells Publishing.
© 2010 by Wells Publishing, Inc. All Rights Reserved. Privacy Policy | Terms and Conditions | Advertise with Us | Contact
Seriously, I have a hard time believing that a credit score drives claims when 80% or more of the claims made are weather related to begin with.
But he does make some very valid points. I wonder why no state insurance dpeartments have used this reasoning to scuttle cerdit scoring. Or is it that insurance comapnies weild too much power?