The Federal Insurance Office's study of auto insurance affordability

The federal government has published one study on whether Americans can afford car insurance: the Treasury Department's Federal Insurance Office (FIO) report of January 2017. Its headline finding was narrow. Of 9,172 ZIP codes where low- and moderate-income residents or minorities are the majority, 845 had an average liability premium above 2% of median household income, and those 845 ZIP codes held more than 18.6 million people. It is not a national verdict on affordability, and as far as we could find, FIO has not published a follow-up.

The study is still worth understanding, because its numbers are quoted often and its definitions are easy to misread. This article explains why FIO did it, how the index works, what it found in the four states it profiled, who criticized it and why, and what it can and cannot tell you about your own premium.

Why a federal office studied auto insurance

Insurance in the United States is regulated mainly by the states (see Why insurance is regulated by states, not Washington), so Washington has no role in approving auto rates. The FIO sits inside Treasury and was created by the Dodd-Frank Wall Street Reform and Consumer Protection Act of 2010. As the January 2017 study explains, Congress tasked FIO with monitoring the extent to which traditionally underserved communities and consumers, minorities, and low- and moderate-income persons have access to affordable insurance products in all lines except health insurance. The report cites 31 U.S.C. 313(c)(1)(B) for that duty.

FIO monitors and reports. It does not set rates or approve forms. Auto insurance was the first line it picked for an affordability study. The report gives its reasons: owning a car is tied to access to jobs, almost every state requires liability insurance to register or drive a car, and stakeholders disagreed about whether coverage was getting more or less affordable for these groups. The report cites an Insurance Research Council release (August 2015) saying affordability was improving for all income groups and a Consumer Federation of America analysis (September 2015) saying there was no increase in affordability for low- and moderate-income households.

The process was public. FIO published a request for information in April 2014, a proposed framework in July 2015, and in July 2016 a notice adopting the final method. The July 13, 2016 Federal Register notice summarizes the comments and explains the choices.

How affordability was defined

The study uses a single ratio, the Affordability Index:

average annual written personal auto liability premium in the voluntary market, divided by median household income, for a ZIP code

FIO presumes liability coverage is affordable in a ZIP code if the index is 2% or less, and unaffordable if it is above 2%.

Why 2%

The Federal Register notice says FIO was most persuaded by the Consumer Expenditure Survey, which showed the average household spending about 2% of its income on automobile insurance (a survey covered in What households spend on car insurance: Consumer Expenditure Survey). The study adds that BLS data showed consumers spending about 1.8% of average after-tax income on vehicle insurance in 2013 and 2015. The notice also points out that other goods have threshold percentages, citing 30% for housing, 9.56% for health coverage and 2% for water and wastewater.

The choice was contested. According to the notice, consumer advocates generally favored the 2% benchmark, while insurers and their trade groups generally opposed a fixed number. The American Insurance Association called it artificial; the Insurance Research Council said no external standard exists to support it; the National Association of Mutual Insurance Companies said it raised the question of how far a number could deviate before coverage counted as unaffordable; and the Property Casualty Insurers Association of America argued the measure favored higher-income groups because low- and moderate-income households by definition spend a higher share of income on necessities. FIO acknowledged "there is not one generally acceptable method or definition of affordability" and adopted 2% as a working benchmark.

Who is counted: "Affected Persons"

The study does not look at everyone. It defines Affected Persons by geography:

• a majority-minority ZIP code, where Black, Native American, Hispanic or Asian American residents exceed 50% of the population;
• a low-income ZIP code, where median family income is below 50% of the metropolitan area median; or
• a moderate-income ZIP code, where median family income is at least 50% but below 80% of the metropolitan area median.

An "AP ZIP code" is one in which these groups are the majority. FIO says ZIP codes serve as proxies, because insurers and statistical agents do not keep the demographic data of individual policyholders.

Why ZIP codes, not counties

Readers sometimes assume the study went by county. It did not. The notice says FIO chose ZIP codes because premium data are collected that way and because ZIP codes show more variation across populations than counties or states, with smaller margins of error than demographic data based on census tracts. Insurers do not sort data by census tract, so tracts were not an option. The study also concedes that tracts would better identify specific communities.

What data went in

The study is more limited than its headline suggests. Per the report's data section:

• It used premium and vehicle counts for policies written at each state's financial responsibility limits, meaning the legal minimum liability limits (see minimum coverage), in the voluntary market, with policy inception dates from July 1, 2014 to June 30, 2015. Residual-market policies (see residual market) were not part of the calculation.
• Premiums covered bodily injury and property damage liability, plus personal injury protection where the state requires it. Uninsured and underinsured motorist coverage were excluded, even though some states require it.
• The data came from states and statistical agents and, in FIO's words, did not include all U.S. auto insurance policies. California and Texas supplied data for different periods, so the tabulated results rest on 2014 to 2015 premium data from the rest.
• ZIP codes with fewer than 100 residents or fewer than 20 insured vehicles at the minimum limits were excluded, as were atypical areas such as military bases.

Two consequences follow. The index measures the cheapest legal policy, not the policy most households carry. FIO noted that many, if not most, consumers bought limits above the minimum, though lower-income and majority-minority ZIP codes were more likely to hold minimum-limit policies. And it measures liability only, so collision and comprehensive, which lenders require on financed cars, are outside the index.

What it found

The headline numbers, from the study's findings section:

All states had AP ZIP codes, and most had one or more with an index above 2%. The state pattern was uneven, and FIO observed that states with more such ZIP codes also tended to have mandatory PIP and higher percentages of uninsured drivers, while cautioning that other factors also matter.

The four states FIO profiled

The report profiled the largest state in each census region (Florida replaced Texas because of data limits):

California stands out for a reason the report spells out: it has a low-cost auto insurance program for income-eligible drivers and, as of the report, barred ZIP code as a rating factor except to measure claim frequency. Florida and New York both require PIP, and Florida also had a high uninsured rate (23.8% in 2012, per the study's source); both show high premium-to-income ratios in many AP ZIP codes.

The report is explicit that these results are best read within a state. Because state laws and markets differ so much, it says the index allows comparison of ZIP codes in the same state but should not be used to compare states, and that it is not appropriate for measuring what any one person pays.

A worked example

Say a ZIP code has a median household income of $45,000, and the average premium for a state-minimum liability policy there is $1,150. These are hypothetical numbers for illustration.

1. Divide the premium by income: $1,150 divided by $45,000 is 2.56%.
2. Compare with the benchmark: 2.56% is above 2%, so the ZIP code would be flagged.
3. Work out the break-even: 2% of $45,000 is $900. The ZIP code would need an average minimum-limit premium of $900 or less to be presumed affordable.

Now change one input. If the same ZIP code's median income were $60,000, the same $1,150 premium would be 1.92% and would pass. This is why the index moves with income as much as with price, and why two neighboring ZIP codes with identical premiums can land on opposite sides of the line.

What the study did not do

A common assumption is that the study proposed policy fixes. It did not: the 2017 study neither recommends nor evaluates any. It lists factors affecting affordability, including state requirements and low-cost programs, uninsured-driver claims, health care costs, fraud, and individual rating factors, and notes that state anti-discrimination laws vary. It describes its work as a baseline for future comparison. Its "next steps" were about data: FIO said it expected to collect more information from large insurers in 2017 through statistical agents so the index could be refined and trends tracked.

It also did not examine:

• Trends. One year of data cannot show whether things got better or worse.
• Full coverage. Collision and comprehensive were not priced.
• Individuals. The index is for ZIP-code groups, not households.
• Causes. It does not explain why a ZIP code is expensive.

Criticism and what happened afterward

When the study came out, the National Association of Mutual Insurance Companies called the ZIP-code approach inherently unreliable and the effort ill-conceived, arguing that the highest-cost areas sat in states such as Florida, Michigan and New York with regulatory and legal problems, according to an Insurance Journal report. The Consumer Federation of America's insurance director said that if government requires drivers to buy insurance, it should make sure prices are affordable.

Follow-up has been thin. In July 2021, 22 consumer and community organizations wrote to FIO asking it to update the 2017 study and produce it annually, as originally proposed, according to a Consumer Federation of America release. In our search through November 2026 we found no second FIO affordability study. If you need a current picture, the substitutes are other measures: the household spending shares in What households spend on car insurance: Consumer Expenditure Survey, state-by-state burden in car insurance affordability by state: premium as a share of income, and uninsured-driver estimates discussed in who goes uninsured and why: cost, lapses and enforcement gaps.

The study's dollar context is also dated. Using BLS data, the CPI index for motor vehicle insurance was about 75% higher in December 2025 than in January 2017, while the all-items CPI was about 33% higher. A 2% test applied to 2014 to 2015 premiums says little about 2026.

How to use it today

• Quote the right claim. "Over 18.6 million people lived in ZIP codes where the average minimum-limit liability premium exceeded 2% of median household income, in FIO's 2017 study, using 2014 to 2015 data" is accurate. "18 million Americans can't afford car insurance" is not.
• Do not compare states with it. FIO says not to.
• Check your own situation. Use your declarations page and the state car insurance requirements lookup to see what your state requires and what you carry. Rules differ by state.

Common questions

Did the Treasury study say car insurance is unaffordable?

It said that in 845 of the 9,172 ZIP codes it classed as majority low- or moderate-income and/or minority, the average minimum-limit liability premium exceeded 2% of median household income. That is a statement about a defined set of places, using a benchmark FIO chose.

Does the index measure what I pay?

No. It uses average premiums for state-minimum liability policies in the voluntary market, not your policy, and it explicitly warns against using it for any one person.

Why was the study limited to low-income and minority areas?

Because Congress's mandate to FIO concerns underserved communities, minorities and low- and moderate-income persons. The report notes it is limited to affordability for those groups.

Has the federal government updated it?

We found no update. FIO said in 2017 it would continue to monitor and report, and consumer groups asked for an update in 2021. A licensed insurance professional or your state insurance department can tell you about state programs aimed at affordability.

Key takeaways

• Treasury's FIO published a single auto affordability study in January 2017, under its Dodd-Frank duty to monitor access to affordable insurance for underserved groups.
• It flagged ZIP codes where the average state-minimum liability premium exceeded 2% of median household income: 845 of 9,172, covering over 18.6 million residents.
• The 2% benchmark, the ZIP-code unit and the liability-only premium were contested choices, and the study warns against state-to-state comparisons or use for individuals.
• The report offered no policy recommendations and rests on 2014 to 2015 data; we found no later FIO edition.
• Treat it as a baseline method, not a current measure, and pair it with newer spending and burden data.