Blacks and Hispanics Pay More for Auto Insurance. Study Tries to Answer Why.

A new study examining potential unintentional bias in auto insurance finds that while Black and Hispanic drivers pay more than other drivers for coverage, they also have more claims and driving infractions.

The study’s authors accept that higher losses and poorer driving records help explain the higher insurance costs for Black and Hispanic drivers and they acknowledge they cannot confidently answer without further research why these drivers have more claims and driving errors.

But the authors also suggest that other rating or policy characteristics, certain discounts, financial status, infrastructure, law enforcement and other factors could be contributing to the higher claims and premiums for Blacks and Hispanics. They argue that these other factors merit future study.

The report (Evaluating Unintentional Bias in Private Passenger Automobile Insurance) published by the District of Columbia Department of Insurance, Securities and Banking (DISB) analyzed data from insurers on one-vehicle/one driver private passenger auto policies in D.C— a group that represents 64% of all auto policies in the District. The researchers also interviewed industry and consumer representatives, and subject matter experts.

“This investigation is a key step toward understanding and addressing financial barriers that could unfairly have an impact on certain racial groups. Our goal is to ensure that D.C. residents are treated equitably, regardless of race, and that insurance premiums reflect actual driving risk,” commented DISB Commissioner Karima Woods.

The insurance industry was quick to claim that the study confirms that auto insurance pricing is not unfairly discriminatory and not based on race.

Premium Gap

According to the analysis, Black drivers pay, on average, 1.46 times more than white drivers, equating to an average annual premium gap of $326. Hispanic drivers pay 1.20 times more than white drivers, and API (Asian-Pacific Islander) drivers pay the same as white drivers.

In dollar terms, that translates into an average annual premium of $705 for white drivers, $1,031 for Black drivers, $849 for Hispanic drivers, and $722 for API drivers.

DISB sought to identify factors that might explain the average premium gaps, and specifically the Black/white average premium gap. DISB looked at age, type of policy, driving record, claim history, gender, and other factors to determine if and how those factors impact premiums.

The analysts looked into cumulative paid losses by race, and found that Black drivers, as a group, represented more claims on average than white, Hispanic, or API drivers. Black drivers’ average losses were 2.38 times those of white drivers. Thus, actuarially, Black drivers are more costly as a group than the other groups, according to the study.

Driving record, which the study says is a “practical and reasonable criterion” because it has a rational relationship to risk, explains some of the Black/white average premium gap. Black drivers have more at-fault accidents and the gap is highest for those with a driving under the influence (DUI) in their record.

But even after accounting for all of these factors, there was still an average premium gap of $271 between Black and white drivers.

Correlation With Race

The authors posit that given that insurers’ rate plans do not vary by applicant race, if the results differ on average by race, then so must the rating characteristics. “Put another way, there must be some factors and/or steps in rating plans that correlate with race in a way that produces different average premiums by race for D.C. drivers,” they wrote.

“What explains the average premium difference by race? There is no evidence that Black and Hispanic drivers are riskier while operating a vehicle or more accident prone. However, there are a variety of factors that may explain this premium disparity including credit-based insurance scores and homeownership discounts. Additionally, some insurers price auto premiums using predictive models. Since premiums are determined mechanically, there may be some factors and/or steps in those models that correlate with race in a way that produces different average premiums by race for D.C. drivers,” the study says.

Examples of such factors that are used by insurers but were not analyzed include credit-based insurance scores; discounts for homeownership, education level, or multiline; payment modes; and telematics.

The report calls these factors “important differentiators for the makeup of premiums,” while questioning the use of those where there are “arguably no rational explanations for why they predict accident risk; or where “there is evidence they are correlated with race in ways that would widen the Black/white average premium gap.”

Multiline discounts, for example, reward customers who have more assets to insure. Also discounts for payment-in-full benefit those who have the ability to pay in advance. This could mean white drivers being asked to pay on average $705 in advance, whereas Black drivers are being asked to pay on average $1,031 in advance.

Why More Claims?

The authors call for additional study on the types and causes of claims by Black and Hispanic drivers to see if infrastructure or other changes may help reduce the claim differential. Additionally they call for investigating why Black and Hispanic drivers are more likely to have driving infractions, and if this might be due to differences in enforcement rather than differences in driving practices.

The analysts further maintain that some factors that affect costs are also ones that favor wealthy policyholders. For example, wealthy drivers are better able to directly pay for some insured losses without filing claims, are more likely to have secure off-street parking, and more prone to live in neighborhoods with better road conditions.

A previous market conduct examination by DISB that looked at specific insurers’ models used for auto insurance concluded the models were “working as intended and did not include any factors that would directly introduce bias into the rating process.” However, that review could not determine if unintentional bias was present.

The authors of this new report on unintended bias say that additional study is warranted into the types and causes of the significantly higher losses for Black and Hispanic insureds to see if infrastructure or other changes may help reduce the claim differential.

DISB said it is exploring additional studies and hopes to develop a “balancing test” to evaluate whether certain factors should continue to be allowed in determining premiums.

Insurer Reaction

Insurer trade groups reacted to the study by claiming it demonstrates that auto insurance rates are based on risk criteria and losses and not on race. They also questioned how DISB determined race in its analysis.

“It is critically important to understand that insurers do not collect race information and do not use race information in any way, including to set insurance premiums,” the American Property Casualty Insurance Association (APCI) said in a statement. “While DISB created and used racial categories in its report, insurers base rates on risk, not race. DISB’s analysis found that, whenever a group paid higher premiums, it also had more accidents, more claims, and generated more in losses, and therefore this group paid more in premiums. Higher losses and more claims translate into higher premiums.”

The National Association of Mutual Insurance Companies stated that insurers do not use the race of a policyholder as a factor in offering or pricing coverage and questioned the study’s methodology. “Even with flawed methodology, the study simply does not support the idea that groups in protected classes are subject to unfair discrimination,” NAMIC said.

Because the data it used from insurers had no information on race, DISB inferred races using a geocoding methodology leveraging US Census data and names and addresses. It is a methodology used by regulatory agencies including the Consumer Financial Protection Bureau when conducting analyses.