Auto Insurance

What Might a Concerned Regulator Do About Systemic and Unintentional Biases in Insurance Markets? Collect and Test the Data

For a long time, consumers (and groups like the Consumer Federation of America) have been raising alarms about insurance company practices that disproportionately harm Black and Brown drivers around the country. Consumers have a right to be treated without bias or unfair discrimination in every market. Still, the need is particularly acute for auto insurance, since virtually every driver in the nation is required to buy coverage from insurance companies. Our research and others’ work in recent years have shown several ways in which insurance companies and the legacy of structural racism in the US leave people of color with worse outcomes – whether it comes to access to the highest quality coverages, the premiums that are charged, or the claims handling process – than would be expected of a market free from unfair discrimination.

However, the industry has long denied even the possibility that institutional bias affects insurance markets and has fought reform efforts at every turn, usually claiming that (notwithstanding CFA’s research) there is no evidence of racially biased outcomes in the insurance market. And that leads us to the important work that is percolating in a few insurance departments around the country (and the need for other state regulators to start testing these insurance industry arguments).

The District of Columbia Department of Insurance, Securities, and Banking is expected to issue a data call requiring insurance companies to provide information on their underwriting, rating, and claims payments that could shed light on the nature and extent of unintentional bias in the auto insurance market. The Department is working with the algorithm auditors at ORCAA, founded by Weapons of Math Destruction author Cathy O’Neil, to conduct the review.

The call will collect information on the following topics:

  • Insurance quotes—whether certain people or groups of people are being quoted higher prices
  • Underwriting decisions—whether certain groups are more likely to be rejected for coverage, or given more expensive insurance options
  • Premiums—whether certain people or groups are paying higher premiums
  • Loss ratio—whether certain groups are getting unfairly charged higher premiums compared to the insurance losses they sustain

Why should consumers care about this? Because discrimination and racism are rampant in auto insurance, but so far auto insurance companies have fought hard against transparency and accountability. For years consumer advocates and racial justice groups have called for investigations of bias and discrimination in insurance. Consumer Federation of America’s investigations have found considerable bias in auto insurance premiums. Moreover, bias doesn’t have to be explicit to harm consumers. An algorithm or practice can be unintentionally biased if it uses bad data, which can result in Black consumers paying higher premiums or being unable to get affordable insurance at all. And the Department’s data call will not only allow us to look at the whole auto insurance market in Washington D.C., but determine how serious the abuse is and decide what to do about it.

While CFA offered some important suggestions for improving the data call, we applaud the Department’s efforts to collect data on unintentional bias in auto insurance, and to determine how to reduce this bias. This data call is a sign that the District of Columbia insurance regulators, like the Colorado Division of Insurance which is developing rules for bias testing of the state’s insurance companies, are taking on the challenge of rooting out the legacy of structural racism and other institutional biases from the insurance market.

Consumer Federation of America submitted comments on how the D.C. data call can be made even better. First, the Department could use the standardized data request that the National Association of Insurance Commissioners developed to collect information, which will make the process simpler and more familiar to the insurance companies when they report data. Second, the Department should collect information about various socioeconomic factors that auto insurance companies use in their business models that (according to CFA’s research) often lead to unfair discrimination against certain classes of consumers. These factors include:

  • Someone’s education level (whether they graduated from high school or college);
  • Employment status and their job/occupation;
  • Homeownership status (whether they own or rent their home);
  • Credit information (auto insurers charge people much higher premiums to safe drivers who have fair or poor credit) ;
  • Whether their car was purchased new or used; and
  • Whether they previously bought minimum insurance coverage or a more robust policy, among others.

Additionally, the Department should collect information about whether insurance claims were flagged for further investigations due to potential fraud, and whether that results in unfair bias.

Finally, we urged the Department to ask insurance companies if any of their divisions maintain information about the race, ethnicity, or national origin of their consumers, and if they do have that information, to provide it. The Department is using an analysis known as BISG (Bayesian Improved Surname and Geocoding) to match demographic data to the insurance data, which is considered an effective and useful way to conduct this type of analysis. However, if insurers – for example, in their marketing divisions – have precise data on the race of their customers, it would allow for an even more precise analysis of the biases in their pricing and practices.

We’ll continue advocating in support of the D.C. and Colorado regulators’ efforts to identify and reform the systems that lead to or exhibit unfair discrimination and bias in the insurance industry. But American consumers need more than two regulators working to fix this problem; there are 49 other state insurance commissioners that need to be challenging insurers in their states to eliminate the unfair and costly impacts of both internal and systemic biases in the insurance system.