Insurance credit scoring plagued with fundamental problems

November 5, 2006

A fundamental problem exists with insurance scoring. The problem is that insurance scores and credit scores reflect correlation, but correlation does not necessarily indicate dependency upon which effective decisions can be made.

For example, there may be a relationship between headaches and brain tumors, but headaches don’t cause brain tumors. While brain tumors may cause headaches, you should not conclude that because a person has headaches that the person has a brain tumor, because there are many other causes of headaches.

With the use of credits scores in the insurance industry, the problem is that there is a correlation between credit scores/insurance scores and the frequency and severity of losses. People with low credit scores have greater frequency and severity of losses than people with high credit scores. But do the factors that result in low credit scores also result in more losses? Or, is there just a correlation and no causation?

Harm occurs to people who have poor credit scores unrelated to the likelihood of the frequency and severity of losses generated by the actual hazards they face. For example, a person who has high medical bills due to a child having cancer may have a poor credit score. However, that has no bearing on whether the child’s parent’s house is more likely to burn down than the neighbor’s who doesn’t have a child with cancer. Yet a poor credit score means they will likely pay much higher homeowners rates.

Credit scores and responsibility

The Insurance Information Institute’s “White Paper: The Use of Credit Information in Personal Lines Insurance Underwriting” discusses the correlation between insurance scoring and higher loss ratios. But it provides nothing other than a suggestion that the reason for higher loss ratios among those with lower credits scores is due to lower responsibility of those with lower credit scores.

A Tillinghast Towers-Perin study was cited in the White Paper, stating: “Tillinghast concluded that it was ‘very unlikely’ that insurance scores and loss ratio relativities are not correlated. … A layman’s interpretation of this result could be that it is very likely there is a correlation between Insurance Bureau Scores and loss ratio relativities.”

The paper also states: “Responsibility is a personality trait that carries over into many aspects of a person’s life. It is intuitive and reasonable to believe that the responsibility required to prudently manage one’s finances is associated with other types of responsible and prudent behaviors, for example: Proper maintenance of homes and vehicles, safe operation of cars. … It is intuitive and reasonable to believe that financially stable individuals are likely to exhibit stability in many other aspects of their lives.”

The contention that credit scores reflect responsibility in other areas than prompt payment of bills, while “reasonable and intuitive” to some, is unsupported by facts. One of the problems with insurance scoring is that a person isn’t told what he needs to improve to raise a score, yet insurers cite insurance scores for nearly everything.

An insurer will cite a poor insurance score as a reason for refusing to add an MCS90 endorsement, even if the insured is well-financed. A case recently was denied where an insured with about a $2,000 annual package premium was non-renewed due to a poor insurance score. The potential insured was not non-renewed due to claims, although it had four claims totaling more than $400,000 in the past three years. The insurer involved also wrote the insured’s $1,000 workers’ comp policy and didn’t non-renew that policy for poor insurance score, although there is a rule against writing unsupported workers’ compensation in California.

Those examples involve insurance scoring in commercial insurance, but I suspect that there are just as many problems in personal lines. Studies done to support credit scores take similar rating structures and separate risks by claims versus premiums within a particular insurer. If you have two preferred homeowners’ policies, one for a person who has a poor credit score and another for a person with a good credit score, there may be a stronger likelihood that the one with the poor score will have a higher loss ratio if you consider no other factors. However, the person with the poor score may live in an older home that has not been upgraded.

Is the low credit score the reason for greater likelihood of loss or is the older home the reason? To demonstrate that poor credit scores are indicators of poor claims performance, you would need identical insurance exposure except for credit scoring.

One of the studies cited in the I.I.I. White Paper tried to address whether credit scores adversely affect people based upon race and involved thousands of drivers from Alaska. It gathered data on “age, gender, residential ZIP code, policy inception date and credit scores and/or rate classifications. About 1,000 of each firm’s policyholders were contacted by phone and asked information on ethnicity, marital status, income level and details of experience if cancelled.”

The paper then stated: “The results of the study were inconclusive, leaving the author to conclude that: ‘while there are statistically detectable patterns in demographics of credit scoring, most of the variation among individual scores is due to random chance or other facts not in this data.'”

Factors for setting premiums

It is acknowledged by those that promote use of credit scores that those with lower credit scores end up paying higher premiums. Do the same insurers that write preferred homeowners write distressed homeowners?

Those insurers that use credit scores for such risks are double dipping. If an insurer has already taken an underwriting action due to age of home or driving record such as raising the price due to a higher exposure, adding premium due to credit scores is double dipping. If the insurer has already non-renewed poor risks due to age of home, driving record or other factors, charging those that don’t have those adverse exposures more based on a poor score is unfair.

The central problem that insurers face is that the data they have upon which to base underwriting decisions is incomplete. Some tickets and accidents are missed in rating auto insurance. Losses that occur may not be reported. Even with accurate data, the data may not be significant enough to accurately predict the likelihood of loss to properly price insurance.

There are many significant factors that drive loss ratio experience. Those familiar with auto insurance know that driving record, years of driving experience, age, marital status, distance to work, total miles, good student status, and performance of vehicle are factors that are used by the insurance industry. But other factors can skew the calculations of insurers. Do they drive in an area prone to dense fog? Do they typically drive while sleepy?

There are probably 20-plus factors that are significant in determining auto rates. Assuming those factors vary from 2 to 10 percent each and that several are impossible, difficult or expensive to determine, insurers look for inexpensive proxies to try to figure out which risks should be written and at what price.

Look at the insurance industry’s premise of “why insurance scoring works.” Quoting again from the white paper: “The statistical correlation between good credit and relatively low insurance loss confirms the reasonable assumption that the responsibility required to prudently manage one’s finances is associated with other types of responsible and prudent behaviors, such as proper maintenance of homes and autos, and safe operation of cars.”

It sounds good, but do a reality check. Assuming you have a good credit score, answer the following:

•Do you always obey posted speed limits?

•Do you consistently avoid tailgating?

•Do you never use the cell phone when driving?

•Do you never eat or smoke while driving?

•Do you annually change the smoke alarm in my home?

•Do you always make sure that there is no brush or debris next to my home?

•Do you always lock flammable liquids in metal cabinets?

•Do you always dispose of oily rags in metal containers?

Most people would agree that these are things responsible people do. But when you consider the “assumption” that good credit scores reflect responsibility in other areas than paying bills on time, set your car’s cruise control to 65. Is it reasonable to think that all the people who pass you will be driven by people with poor credit scores?

Credit scores are not a proxy for responsibility. Credits scores measure the timeliness of payments made on bills and total debt level. That applies to all types of bills. Medical bills alone severely impact credit scores. Recent studies have shown about half of all bankruptcies in the United States are related to medical expenses and untold numbers of others who haven’t filed bankruptcy are struggling under debt caused by medical bills.

Some states, including Texas, have prohibited insurers from taking action on a low score if it is due to medical expenses. If you eliminate one cause after another for poor credit scores as factors for insurance rating, you can see that there are serious questions about a causal relationship between low credit scores’ “irresponsible behaviors” that result in higher auto and homeowners losses.

Casual and causal relationships

Despite the lack of demonstrated causation, insurers use credit scores as a factor in rating and underwriting. There is speculation that people who don’t pay their bills on time are more likely to have other problems that increase the likelihood of claims. For example, dishonest people may not pay their bills on time and a dishonest person may be more likely to file a fraudulent insurance claim. There is also speculation that people who don’t pay their bills on time aren’t as likely to maintain their property and that can lead to claims.

Yet I have never heard of a study that shows the people with bad credit scores don’t fix their car’s brakes or replace the batteries in their smoke alarm, or any other actions or inactions that make any difference in the likelihood of loss as compared to persons with higher credit scores. Somewhere along the line, people have blurred the meanings of “casual relationship” and “causal relationship.” Because some event happens after another event doesn’t mean that the two events are related to each other in a cause and effect relationship.

“Causal” relationships are the key to the business of insurance, not “casual” ones. What is the causative connection between low credit scores and higher loss ratios?

If causation doesn’t exist in credit scoring, it means that people, through no fault of their own and although there is no increased hazard at all, are either paying higher premiums than they should or they are being denied insurance altogether. Nevertheless, it costs less for insurers to use credit scores than it costs to conduct physical inspections and have human underwriters and loss control engineers gathering and analyzing data to determine the actual hazards that increase the frequency and severity of losses.

If insurers can reduce or eliminate underwriters’ salaries by letting a computer decide whether to decline or uprate risks based upon a credit score, think of the money that is saved. If a loss control engineer can be eliminated by having a declination decided by a computer based upon a credit score, it will save even more.

Some insurers combine some other underwriting factors with credit scores and call them “insurance scores.” A computer weights the factors and the insurer doesn’t even tell its underwriters what those factors are. Then, when someone is denied insurance, uprated or cancelled because of an “insurance score,” the underwriter can’t tell the customer why they were denied other than that they have a failing score. Thus, an insurance customer is prevented from getting valid information from the insurer about why they have been denied insurance.

When insurers use credit scores, whether insurers intend to do so or not, insurance consumers are treated like they don’t matter. Insurers will sometimes acknowledge that people get harmed along the way, but overall, if an insurer has better loss experience than if it doesn’t use credit scores, it will choose to use the scores.

Some view the harm as disappointing, but feel that is a cost built into the system. It’s easy to slough off harm when it only affects other people. But when such things happen to you, you will see the injustice.

Thomas E. Nelson, president of Hallmark Insurance Associates Inc. in Fresno, Calif., has worked in the retail agency business since 1970. E-mail: thomasenelson @hallmarkins.net. Phone: 800-274-7501.

Was this article valuable?

Here are more articles you may enjoy.

From This Issue

Insurance Journal West

Insurance Journal Magazine