ZIP-code pricing and racial disparities: what investigations found

Several investigations since 2015 have found that drivers in predominantly Black, Hispanic or otherwise minority ZIP codes are quoted or charged more for auto insurance than drivers elsewhere, in some cases by large margins even after accounting for income, density or, in the best-designed work, local claim payouts. What none of them has established is why. Insurers answer that they do not collect race and price by expected loss, regulators have disputed some methods, and the one regulator that ran its own review found a premium gap but also a claims gap.

This article walks through what each investigation did, what it showed, what the defense says and what is still unknown. It is a summary of published work, not a legal conclusion about any company.

How a ZIP code gets into your premium

Most auto insurers rate by territory: the area where a car is garaged. The insurer groups ZIP codes (or smaller areas) into territories, estimates the claim cost in each, and applies a multiplier to the base rate. Territory is one rating factor among many; Why your ZIP code changes your car insurance price covers the mechanics and territory rating defines the term.

The legal setting differs by state. In California, for example, the rating regulation lists driving record, annual miles driven and years of driving experience as the mandatory factors. Territory is an optional factor, expressed through relative claim frequency and severity, with no more than twenty categories per factor, formed by grouping ZIP codes with similar average claims, according to the state's regulation, 10 CCR 2632.5 (as published by Cornell's Legal Information Institute). Other states allow more discretion, and some states limit how heavily territory can count.

The reason territory exists is real: claim costs differ by place. Traffic density, theft, road conditions, weather, medical costs and litigation all vary. The question these investigations ask is whether the territory multipliers go beyond what those costs justify, and whether a factor that reflects place also reflects race.

2015: Consumer Federation of America and the quote comparison

In November 2015, the Consumer Federation of America reported that quotes from the five largest insurers averaged $1,060 in predominantly African American ZIP codes and $622 in predominantly white ZIP codes, a gap of 70%. The comparison controlled for population density and income level: in the densest urban centers the gap was 60% ($1,797 versus $1,126), in rural ZIP codes 24% ($669 versus $542), and in upper-middle-income ZIP codes 194% ($2,113 versus $717).

The data were quotes bought from Quadrant Information Services for one driver: a 30-year-old woman with a perfect driving record who rents her home, works a clerical job and has a fair credit score. They covered basic liability only. The organization said it was "not rushing to judgment about why this happens" and called for regulators and the industry to look harder.

The strengths of the work are its national scope, its disclosure of data source and profile, and its attempt to hold density and income constant. Its limits are those of any quote study: quotes are not premiums actually paid, one profile cannot stand for a whole market, and quotes alone say nothing about the claims costs behind them. Our companion article reading auto insurance studies by who paid for them explains why that last limit matters.

2017: ProPublica and Consumer Reports add payout data

In April 2017, ProPublica and Consumer Reports published what they called the first use of industry payout data to measure racial disparities in car insurance premiums across states. It covered California, Illinois, Texas and Missouri, the only four states that, ProPublica said, provided ZIP-code-level data on liability claim payouts when it requested records from all 50 states and the District of Columbia.

Method.

• More than 100,000 liability premiums, supplied by Quadrant, for one customer type: a 30-year-old woman with a safe driving record.
• Those premiums were compared with the average amounts insurers paid in liability claims in each ZIP code, from state data covering five years in California, Texas and Missouri and three in Illinois.
• A "minority" ZIP code was one with more than 66% non-white population in California and Texas, and more than 50% in Missouri and Illinois, chosen to keep the sample large enough.

Findings as reported.

• In Illinois, 33 of the 34 companies analyzed charged at least 10% more, on average, for the same safe driver in minority ZIP codes than in comparably risky white ZIP codes; six, including Allstate, averaged more than 30%.
• In Missouri and Texas, at least half of the insurers studied charged more for a safe driver in high-risk minority communities than in comparably risky non-minority ones.
• In California, eight insurers charged more than 10% above in risky minority ZIP codes compared with similarly risky ZIP codes where more residents were white.
• Households in minority-majority ZIP codes spent about 11% of income on auto insurance, ProPublica's analysis of Census data and 30 million quotes found, against about 5% in majority-white ones, with the U.S. Treasury's affordability benchmark being 2% or less.

The article was explicit about uncertainty: "It isn't completely clear why some major auto insurers persist in treating minority neighborhoods differently." It offered two possibilities, a vestige of past practices and proprietary algorithms that may inadvertently favor white neighborhoods, and called both possibilities rather than findings.

What insurers and regulators said back

The replies are part of the record and deserve the same care as the findings.

• The Insurance Information Institute said insurers "do not collect any information regarding the race or ethnicity of the people they sell policies to" and "do not discriminate on the basis of race."
• USAA said some areas may have higher rates due to congestion that leads to more accidents or to higher crime that leads to more auto theft.
• Nationwide said it uses "nondiscriminatory rating factors in compliance with each state's ratemaking laws."
• The California Department of Insurance objected that an individual insurer's losses in a ZIP code may differ significantly from the industry average, and called the study's methodology flawed.
• The Illinois Department of Insurance called the methodology "incomplete" and said it "oversimplifies" the comparison.
• The Texas Department of Insurance said it was reviewing the analysis; the Missouri department did not respond to ProPublica's inquiries.

ProPublica acknowledged the central limit directly: insurers do not release their own losses by ZIP code, so the analysis used industry-wide totals, covering almost all insurers in California, Illinois and Missouri and 70% of insurers in Texas. It said it was possible that some insurers had proprietary data justifying their prices. In a check using Nationwide's own California losses, it reported finding a larger disparity (21%) than with industry data (14%).

The Washington, D.C. review: a regulator measures

The most direct regulator study so far came from the District of Columbia's Department of Insurance, Securities and Banking. In a November 2024 release, DISB reported that it had examined potential unintentional bias in private passenger auto premiums for Black and Hispanic drivers. Using inferred race, it found that Black drivers paid on average 1.46 times as much as white drivers, an average annual gap of $326, and Hispanic drivers 1.20 times as much. Drivers it classified as Asian-Pacific Islander paid the same as white drivers.

Two details matter for interpretation. DISB said Black and Hispanic drivers also incurred claims with an even higher difference than white drivers, and said this points to a need for further study of why. It also said that while Black drivers filed more claims and incurred higher losses on average, the differences "may reflect broader systemic factors rather than inherent driving risk," and that certain underwriting factors, including credit scores and homeownership, appear correlated with race and may indirectly affect premium disparities.

The Consumer Federation of America welcomed the findings and pointed to credit history as a likely driver. Notice what each party emphasized: the advocate stressed the premium gap, and the regulator's release stressed the claims gap as well. Both are in the release; neither cancels the other. See credit-based insurance scores and car insurance for how credit enters rates where it is allowed.

What these studies can and cannot show

The actuarial defense, and why disparate impact complicates it

The traditional rule in rate regulation is that rates must not be "excessive, inadequate or unfairly discriminatory." ProPublica reports that the model language recommended by the NAIC defines discrimination as price differentials that "fail to reflect equitably the differences in expected losses and expenses." On that cost-based reading, a territory factor is lawful when it tracks expected claims, even if it correlates with race.

Actuaries have defended that standard. A 2009 paper by Michael J. Miller in the Casualty Actuarial Society's E-Forum, summarized on the society's site, argues that historical sources define unfair discrimination in terms of cost, that disparate impact is a different concept unrelated to cost, and that adopting it as the test would weaken risk assessment.

The difference between the two ideas is the heart of the policy dispute:

• Intent (disparate treatment). The insurer uses race, or uses a factor in order to affect people by race. Insurers deny this, and nothing above contradicts them.
• Effect (disparate impact). A neutral-looking factor, such as a ZIP code, produces outcomes that fall more heavily on a protected group. The question becomes whether the factor is justified by cost and whether a less discriminatory alternative exists.
• Proxy. The factor stands in for something else, like income or race, because it correlates with it. The DISB release raises this about credit and homeownership.

Whether disparate impact is a legal test in auto rate regulation depends on the state's statutes and regulators, and this article does not state a rule for any state. Where the question matters to you, a licensed attorney or your state insurance department can address it.

The insurers' statement that they do not collect race is also worth reading precisely. It speaks to direct use. It does not answer a proxy argument, which asks about effects of factors that are collected. And it works both ways: regulators who want to test for bias must infer race, as DISB did, because the data are not in the file.

How regulators and legislators have responded

The responses vary, and the following are examples rather than a survey.

• Capping territory. Connecticut's Public Act 10-7 (effective January 1, 2011) requires rating plans for private passenger nonfleet auto insurance with territorial classifications to assign a weight of 75% to the territory's own loss cost indication; the remaining weight goes to the statewide average, according to the act text. The Consumer Federation of America cites that law as the model for a Massachusetts bill, S.820, that it supported in testimony in May 2025. Check the legislature's site for that bill's status.
• City-wide territories. ProPublica reports that Illinois in 1972 required a single territorial rate within Chicago for bodily injury coverage, after complaints about four Chicago rating territories, but that the rule does not cover property damage coverage, where rates may still vary by neighborhood.
• Stricter limits on ZIP code. New York's 2026-27 budget includes limits on ZIP-code rating. Trade press reports describe the measure differently: Insurance Journal said in May 2026 that insurers would be prevented from using ZIP codes, education level and occupation, while other accounts describe a limit on ZIP code as the primary basis. Read the statute and the Department of Financial Services guidance before relying on either description. See What drives car insurance costs in New York for the state's background.
• Data collection. Regulators in a few states already gather ZIP-level loss data; ProPublica found only four that did in 2017. The NAIC created a special committee on race and insurance in July 2020, which, according to the NAIC, was disbanded in 2024 with its work folded into the lettered committees.

Other rating factors have also been restricted on fairness grounds in some states; rating factors some states have banned: gender, credit, occupation lists them.

What remains uncertain

• Insurer-level answers. Without each company's own ZIP-level losses, any outside study works from industry averages, and insurers and regulators can fairly say a given insurer's experience may differ. More states collecting data would reduce that dispute.
• Current practice. Rating plans change yearly. Studies from 2015 and 2017 describe the markets of those years.
• Causes. The studies do not separate territory from credit, vehicle value, coverage choices, shopping behavior or distribution channels, and DISB said further study is needed to understand why its differences exist.
• Coverage scope. Most of the work is on basic liability. Collision and comprehensive, which depend on car value and theft, were not tested in the CFA study, and ProPublica's analysis was of liability.
• Remedies. Capping territory narrows differences between places but, as the Connecticut approach itself shows, still permits territory-based pricing; whether it helps the intended drivers more than it shifts cost elsewhere is an empirical question for each state.

What a driver can do with this

1. Shop the same coverage with several insurers. ProPublica's profile of a Los Angeles driver found that switching insurers would save him nearly $400 a year on two cars, and the article shows insurers differing widely for the same address.
2. Ask each insurer what your territory factor is. Request the base rate and territory relativity from the filing, or ask a licensed agent to explain it. how car insurance rates are calculated explains the steps.
3. Report the garaging address accurately. Using a lower-priced address is misrepresentation and can lead to a denied claim or canceled policy; see premium fraud: garaging address and driver misrepresentation.
4. Check your state's tools. Your state insurance department may publish shopper guides or complaint information; start with the state car insurance requirements lookup tool and the regulator's site.
5. Complain if you think a rate is unfair. Most state insurance departments accept consumer complaints and can ask an insurer to explain how it rated a policy.

Common questions

Do insurers ask for my race?

The Insurance Information Institute's chief actuary told ProPublica in 2017 that insurers do not collect race or ethnicity information from policyholders. That is the industry's position; the neighborhood studies in this article classify ZIP codes by their population, and the D.C. review inferred individual drivers' race.

Is territory rating illegal?

No. It is widely used and permitted, though states regulate how it is constructed. California requires territories to be built from groups of ZIP codes with similar claim experience, with a limit on categories; Connecticut limits how heavily territory loss costs may count. The fairness debate is about how far it can go and whether its effects are justified.

Why not just ban ZIP codes?

Claim costs really do differ by place, so removing location from pricing shifts cost from some drivers to others and may make coverage harder to supply in high-loss areas. Advocates answer that the gap between prices and loss differences shows a limit is needed. States are trying different balances, as the Connecticut, Illinois and New York examples show.

Did any of these investigations prove discrimination?

None of them claims to prove intentional discrimination. They document premium gaps by neighborhood demographics, and some compare those gaps with loss data. Whether a gap is unlawful is a question for regulators and courts in each state.

Key takeaways

• Consumer Federation (2015), ProPublica and Consumer Reports (2017) and D.C. regulators (2024) each found higher auto premiums in minority neighborhoods or for minority drivers, using different data and methods.
• The studies establish a pattern for the profiles, years and places tested. They do not establish intent, and they cannot isolate territory from credit and other factors.
• Insurers say they price on expected loss and do not collect race; the California and Illinois regulators challenged ProPublica's method, and D.C.'s own review found a claims gap alongside the premium gap.
• States have responded in different ways, from Connecticut's 75/25 cap on territory weight to Illinois's Chicago-wide bodily injury territory and New York's 2026 budget limits.
• As a driver, compare quotes, keep your garaging address accurate and ask your state insurance department if a rate looks unexplained.