Credit Score Use in Insurance: Actuarial and Regulatory Issues

What Are Credit-Based Insurance Scores?

Credit-based insurance scores are numerical ratings derived from information in a consumer's credit report that are used to predict the likelihood of filing insurance claims. These scores are distinct from traditional credit scores used by lenders. While a lending credit score predicts the probability of defaulting on a loan, an insurance score predicts the probability and expected cost of insurance claims. The models use similar input data (payment history, outstanding debt, length of credit history, new credit inquiries, and types of credit used) but weight the factors differently to optimize for insurance loss prediction.

Insurance scores are most commonly used in personal auto and homeowners insurance pricing. The score is typically one of several rating variables that together determine the policyholder's premium. A consumer with a high insurance score (indicating lower predicted claim costs) will pay less than one with a low score, all else being equal. The magnitude of the impact can be significant, with rate differentials of 50% or more between the best and worst score tiers in some states and companies.

Actuarial Evidence

Numerous actuarial studies have demonstrated that credit-based insurance scores are statistically significant predictors of future insurance claims, even after controlling for traditional rating variables like age, gender, territory, and driving record. The relationship is robust across multiple lines of business, geographic regions, and time periods. Studies commissioned by the SOA, state insurance departments, and the Federal Trade Commission have consistently confirmed the predictive power of insurance scores.

The strength of the correlation is notable. In multivariate models, insurance score often ranks among the top three most predictive variables for personal auto and homeowners loss ratios, alongside driving record and prior insurance history. The relationship appears to reflect underlying behavioral characteristics related to risk awareness, attention to detail, and financial responsibility. Consumers who manage their financial obligations carefully also tend to maintain their property, drive more cautiously, and take steps to prevent losses.

How Insurance Scores Are Built

Insurance scoring models are developed using standard predictive modeling techniques. The actuary or data scientist starts with a large dataset containing both credit report variables and subsequent insurance claim outcomes. The model is trained to identify which credit variables are most predictive of claims and how they should be weighted. Common modeling techniques include logistic regression, decision trees, and proprietary algorithms developed by companies like LexisNexis and TransUnion.

Important credit variables for insurance scoring include payment history (the most heavily weighted factor), outstanding debt relative to available credit (utilization ratio), length of credit history, number of recent credit inquiries, and the mix of credit types (revolving, installment, mortgage). Certain credit events like bankruptcies, collections, and charge-offs are strong negative signals. The models specifically exclude income, employment status, and other factors that are prohibited from use in insurance rating in most jurisdictions.

Regulatory Controversy

Despite strong actuarial support, credit score use in insurance is one of the most contested regulatory issues in personal lines. Critics argue that insurance scores have disparate impact on minority and low-income populations, effectively charging higher premiums to groups that already face economic disadvantage. Research has shown that on average, Black and Hispanic consumers have lower insurance scores than white consumers, and lower-income consumers have lower scores than higher-income consumers. This correlation raises concerns about fairness even if the scores are not directly based on race or income.

Several states have restricted or banned the use of credit in insurance rating. California prohibits its use in auto insurance. Hawaii, Maryland, and Massachusetts have imposed significant limitations. Other states allow credit scores but require insurers to offer re-scoring if a consumer's credit has been affected by extraordinary events like job loss, divorce, or medical emergency. The NAIC has developed model legislation that permits the use of credit information subject to certain consumer protections.

The Fairness Debate

The core of the debate involves balancing actuarial accuracy against social equity concerns. From a pure risk classification standpoint, insurance scores improve pricing accuracy by identifying risk differences that other variables do not capture. Removing credit from the rating algorithm would force low-risk consumers (who tend to have higher insurance scores) to subsidize high-risk consumers (who tend to have lower scores). Proponents argue this cross-subsidization is itself unfair.

Opponents counter that actuarial correlation is a necessary but not sufficient condition for using a rating variable. Just because a variable predicts claims does not mean it is appropriate for pricing. The variable should also have a logical causal relationship to risk, and its use should not produce outcomes that conflict with broader social values. The actuarial profession's standards require actuaries to consider whether a rating variable is a proxy for protected characteristics, even if it is not directly based on those characteristics.

Common Pitfalls

A frequent mistake in implementing insurance scoring is failing to comply with the Fair Credit Reporting Act (FCRA), which requires specific disclosures and adverse action notices when credit information is used to make decisions unfavorable to the consumer. Insurers must notify applicants when a credit-based score results in a higher premium and provide information about how to obtain their credit report. Another pitfall is treating insurance scores as static. Consumers' credit profiles change over time, and regular re-scoring ensures that premiums reflect current risk levels.

Actuaries should also be cautious about model stability. Credit scoring models developed during normal economic conditions may behave differently during recessions or other periods of widespread financial stress. If a large portion of the insured population experiences credit deterioration simultaneously, the model's predictions may become less reliable, and the resulting rate changes could create affordability issues.

Study Tips

For exam preparation, understand the distinction between credit scores used for lending and insurance scores used for rating. Be able to articulate the actuarial evidence supporting the predictive power of insurance scores and the key credit variables used in scoring models. Know the major regulatory positions, from states that permit credit scoring to those that ban it. Be prepared to discuss the fairness debate from multiple perspectives, including the actuarial, consumer advocacy, and regulatory viewpoints. Review the FCRA requirements applicable to insurance use of credit information and understand how extraordinary life events provisions work in states that have adopted them.