What Factors Should Underwriters Consider?

Should underwriters use all sorts of data or restrict themselves to models and loss data?
By: and | August 1, 2013

Point: Use long-term trends in underwriting, not unrelated data.

Using an insured’s loss history in combination with group data and models are the best ways to ensure profitable underwriting.

Underwriting is an art, and the best underwriters know it’s best to filter out unrelated data that could result in an unprofitable decision.

Anyone who has looked incredulously at the latest fad, whether it’s buying $350 ripped jeans or frenzied Beanie Babies collecting, knows that long-term trends are better predictors of behavior.

An insured’s history of losses, in combination with modeling and group data, should be the primary factors in any analysis of risk from an underwriting perspective. History has shown that it’s nearly always useless to try to predict future behavior. Vague gut feelings are frequently wrong, and this is true for underwriters as well as for most individuals.

In fact, underwriting is a perfect example of collective intelligence being able to produce better results than any one individual, as most recently outlined in James Surowiecki’s The Wisdom of Crowds.

In that book’s opening anecdote, he writes of an experiment where the aggregate guesses of a crowd at a county fair, many of whom had no knowledge of the subject, more accurately guessed the weight of an ox than the individual estimates of experts who were asked.

What is underwriting other than a collection of the wisdom of crowds? The collected history of insureds is the best predictor of what the future will hold. Adding in unrelated data that purports to predict behavior creates a flawed calculation.

Even the attempt to add in that data can put carriers in a potentially risky situation. Such is the case when underwriting specific, individual risks butts up against the strict laws against the use of health-related data in workers’ compensation, for example,

In addition, the use of genetic testing results raises serious ethical, and potentially legal, questions if used to underwrite group life-insurance policies.

How can an insurance carrier really calculate the best price if the underwriting is flawed by using unrelated data? New underwriting factors, unrelated to the specific risk should be ignored, especially if the data invades the privacy of the insured.

Using past history of an insured in combination with modeling and group data is the prudent way to analyze risk and underwrite.

Counterpoint: Use all available data in underwriting.

The case can be made that when underwriting fails, it’s a result of a failure to use all available data.

The analysis of seemingly unrelated factors can many times point an underwriter in the right direction.

Nowhere is this more true than in the current situation facing workers’ compensation underwriters. Those who insure the risk are pounding their heads against the wall, facing combined ratios approaching 120, in many cases.

But some would restrict the ability of underwriters to gauge this risk, thinking they are arguing in defense of civil liberties. A patient’s mental health history or whether their father was a heavy drinker has no bearing on the patient’s loss history or the ability to predict a catastrophic workers’ compensation claim, they say.

If a patient has genetic addictive tendencies and suffers from depression, however, those are factors that could influence the outcome of an injury claim. Using those red flags to stage an intervention to prevent that patient from abusing opioids is something that will add to that patient’s quality of life, not subtract from their civil rights.

The use of credit scores in personal lines underwriting is another area that has generated controversy. Opponents say credit scores have no relation to whether someone can operate or maintain a vehicle safely. Really?

Is it that much of a stretch to conclude that someone who can’t bring order to their finances will lack order in other parts of their life? To me, that sort of thinking collides with common sense.

A failure to use all available data at this point in time seems particularly short-sighted. At no other time have we had the analytic tools that we have now. These tools are powerful, but to run optimally they need good data, and lots of it.

I think it’s time we drop the collective, foolish pride that stops us from sharing important information on such topics as family obesity, mental health or financial ineptitude and face those issues together as a society. People already post more on Facebook than I want to know anyway.

Insurance will always be a modifier of behavior, but it can’t modify behavior with one hand tied behind its back, deprived of useful, relevant data.

The late Anne Freedman is former managing editor of Risk & Insurance. Comments or questions about this article can be addressed to [email protected].

More from Risk & Insurance