July 2012

Consumer Insurance Score: Exploring the Possibility of an Alternative to Credit-Based Insurance Scores
View as a Print-friendly PDF

* This article expresses the opinions of the author and is being included in the Newsletter to alert policymakers on emerging trends in rating practices. It is not meant to represent the position or opinions of the NAIC or its members, nor is it the official position of any staff members.

The Nobel Prize physicist Niels Bohr once said,“The opposite of a correct statement is a false statement. But the opposite of a profound truth may well be another profound truth.”

Within the past two decades, the recognition of a statistical correlation between certain elements of personal credit information and insurance claims was viewed by the property/casualty (P/C) insurance industry as nothing less than a “profound truth.” The Fair Isaac Corporation (FICO) and other credit information vendors agreed.

For underwriting purposes, it is most efficient to gather individual risks into large, similar groups. Insurance scores facilitate this process in an efficient manner. However, for an individual, it is important to be evaluated on the individual’s unique characteristics, and therein lays an essential tension.

Pricing insurance products requires evaluating objective factors like expenses, replacement value, etc., as well as subjective factors (such as driver experience, for example). The P/C industry has long sought, and continues to seek, ways to efficiently, objectively and consistently group individual risk characteristics. The industry prefers to use risk-differentiation factors that are predictive of loss, easy to obtain, easy to verify and not easy to manipulate.

Consider that, without a technical approach to evaluating subjective factors, a human underwriter must decide what they believe a driver’s experience is worth. The correct answer should be “it depends.” Most underwriters would want more information prior to answering this question. Did the driver take a formal driver education class? Were the classroom lessons actually employed by the driver? How many hours and miles of actual driving experience have occurred?

Each of these considerations represents a best estimate of value on a continuous scale by the individual underwriter evaluating the risk, based on the information available. If a particular factor scores better than average, how much better and what is it worth? Would another underwriter arrive at the same conclusions?

Dealing with this subjectivity as an underwriting company takes time, costs money and introduces variability through human interpretation of information. Insurance companies invest considerable resources to maintain structured underwriting guidelines to support this human decision process.

Grouping similar risks provides common measures across different “books” of insurance contracts and helps carriers ensure that actual losses are similar to estimated losses.

The ability to translate subjective variables to measurable values, or numeric data, represents an enormous benefit to insurance carriers and explains the industry’s embrace of a method that satisfies such a critical and strategic function. Computers, software and business rules can do more of the underwriting work.

Following the introduction of credit scores for mortgage underwriting in the banking industry, insurance risk scores began showing up in the mid-1990s when the insurance industry started using credit-based insurance scores as a proxy for a risk score in the risk-selection portion of the insurance underwriting process. With a clear correlation to claims, the credit score allowed an automated (or near-automated) classification and grouping of individual risks. Later, the credit-based insurance score was expanded to include the pricing portion of the insurance underwriting process.

Consumer advocates soon surfaced to challenge the practice of using credit scores for underwriting purposes, arguing that credit scores were never intended for insurance underwriting and some consumers were treated arbitrarily, if not unfairly, with this method. Consumer advocates and regulators have expressed concerns that some consumers cannot easily understand how their credit scores are calculated or what any of the scores really mean. That one’s use of credit affects insurance risk selection and pricing is less intuitive to the public than use of credit for lending purposes.

Further, consumers complain that a significant number of credit reports contain inaccurate information. Actions that can lower one’s credit scores, such as closing accounts or having low credit card limits, might matter from a lending perspective, but those actions can also increase the price one pays for their insurance coverage without an intuitive relationship to insurance. As an example of how confusing this can be for consumers, a recent Web search returned several different results with specific, yet contradictory, advice to individuals seeking to improve their credit score.

If all the people in the United States with a credit score of 687, the median Experian credit score at time of this writing, were lined up and we could see and talk to them individually, what differences would we find? Would they all look, think and act identically, or would we likely observe both similarities and differences? It is precisely in the differences we might find that consumer advocates are most interested and want the underwriting process to contemplate.

Consumer advocates have had some limited success in restraining the practice of credit-based insurance scores, and they have anxiously watched as study after study affirms the correlation between credit score and claims, but not the cause.

The P/C industry has successfully defended credit-based insurance scoring by showing actuarial evidence of strong correlation, continuing with at least some level of regulatory approval in all but four states. The fundamental position of the insurance industry is that credit scores are predictive and, therefore, honor the industry’s social contract to charge premiums based on actual cost of risk.

Advancing beyond risk selection, the P/C industry has begun using credit scores to calculate premium. In spite of legal requirements of transparency through filing of rate calculation formulae, regulators and consumers want to better understand insurance-pricing techniques. The complexity of the formulas, how insurers use them to assign a person to a rating tier and the interaction of credit variables with other risk classification variables makes it difficult for regulators and consumers to fully understand how the credit information impacts prices charged to consumers with otherwise similar risk profiles.

At a time of strong policyholder surplus in the well-capitalized P/C industry, American consumers are facing huge economic hurdles in their homes. In the present economy, even professional, middle-class individuals have been adversely impacted by outsourcing, offshoring and staff reductions. If you’re reading this, you have probably witnessed some of these setbacks firsthand, as insurance industry employment has not been immune. Does a shift in credit scores cause an equal shift in claims?

The P/C industry serves an important function in society. Without it, we would never be able to enjoy the quality of life we do in modern civilization. As in any business, a reasonable profit must be possible for investors and owners of insurance companies to be willing to put their capital at risk to perform this service to society. Unlike many businesses, insurance companies are regulated and are required to meet minimum solvency standards. The other part of the bargain for the insurance industry is to be good corporate citizens by making coverage available and spreading costs fairly. This requires them to treat policyholders and claimants fairly and consistently.

Ken Blanchard wrote in his book Raving Fans, “Your customers are only satisfied because their expectations are so low and no one else is doing any better.” With the honorable mission, history and heritage of the insurance industry, I suggest to you that this is not good enough!

In 2007, as part of the rationale for endorsing the use of credit scores for insurance underwriting, the Federal Trade Commission (FTC) concluded, “No readily available alternative model exists.” So, let’s take a walk.

What if there could be an alternative model? What if that model was based on information that is voluntary, verifiable, objective and at least as predictive as credit scores? What if that model was equally predictive for appropriate credits (discounts) as it was for debits (surcharges)? Visualize the alternative model based on common and ubiquitous data, external to any individual rating schema, so consumers and prospective carriers would be calibrated to a standardized risk score. This model would improve the ability for consumers and small businesses to compare coverage offers and for carriers to compete.

As technology improves and more information can be harnessed for predictive modeling purposes, a much clearer picture of an individual risk is possible. The implications of such a model are profound, including underwriting efficiency, industry growth, regulatory transparency and, perhaps most important, consumer efficacy and goodwill.

However, little incentive presently exists within the insurance industry and its vendors to rethink a business model that is generating surplus in spite of continued catastrophe losses and weak investment income. Any industry resources capable of designing and implementing predictive models are fully engaged with other strategic work.

According to some, the issue is urgent. If an alternative model is desirable and feasible, who has the motivation and capability to build it? Should the aim be to replace the credit score for underwriting or merely to offer an alternative, objectively calibrated picture of individual risk characteristics?

The profound truth of credit score correlation to claims is not disputed. The plausible suggestion that another equally profound truth is available begs a closer look.

Perhaps the timing is right to consider the broad implications of the present state and explore the potential for an alternative consumer-risk model designed specifically for insurance-underwriting purposes.

Although every question has not been answered, we know much about what can be done with the data and technology available. Wouldn’t it be nice to finally get rid of all the noise created by the use of credit-based insurance scores?

Among the alternatives to credit based risk scores, telematics offer the greatest hope for immediate change in pricing auto insurance. Only with understanding comes acceptance. Use of telematics has the advantages of being both objective and more understandable to the public than credit-based insurance scores.

To facilitate such a large change, use of data from telematic devices can be voluntary while the technology matures and capabilities are refined.

Over time, the information gleaned from telematics devices on cars could deliver breakthroughs in risk classification. Imagine a consumer-risk dashboard, providing real-time feedback on risk factors and behaviors within the insured’s control. Everybody wins.

Whether you are an insurance professional, regulator, vendor or consumer advocate, this is your invitation to join the discussion and shape what happens next.

* About the author: Steven A. Bennett is an independent organizational innovation consultant with more than 25 years’ experience solving business and technical problems in financial services, government, non-profit and other industries.

Copyright © 2012 NAIC, All rights reserved.