Blog
What households spend on car insurance: Consumer Expenditure Survey
How the BLS Consumer Expenditure Survey measures vehicle insurance spending by income, age and region, and why it differs from NAIC averages.
In the Consumer Expenditure Survey's 2025 tables, released on October 29, 2026, the average U.S. household spent $2,153 on vehicle insurance, about 1.9% of its income before taxes by our calculation. The share is far from even: the lowest-income fifth of households spent $1,011, roughly 5.6% of their income, while the highest-income fifth spent $3,147, roughly 1.1%. Spending also peaks for households headed by someone aged 45 to 54 and is higher in the South and West than in the Northeast and Midwest.
These are spending numbers, not prices. They count what households report paying, averaged over everyone, including households with no vehicle. That is why they rarely agree with the NAIC's average per insured vehicle or with a quoted rate, and why this survey is most useful for questions about budgets rather than about what a policy should cost.
What the survey is and who runs it
The Consumer Expenditure Surveys (CE) are collected by the Census Bureau for the Bureau of Labor Statistics (BLS). BLS's CE home page describes two parts: an Interview Survey that covers major and recurring items, and a Diary Survey that covers smaller or more frequently bought items. Published tables combine both.
The CE exists mainly to describe how Americans spend money and to supply the weights for the Consumer Price Index. BLS's Handbook of Methods says annual CE data are used for the CPI weights and that these weights are updated annually. So the survey you are reading about here is the reason motor vehicle insurance carries a particular weight in the CPI, a link explored in The CPI for motor vehicle insurance: how price change is measured.
The unit of observation is not the household but the "consumer unit": per BLS's tables guide, a group of people who live together and share major expenses, which differs from a household because some people living together are financially independent.
How the data are collected
The survey design, as described in the BLS design chapter:
- Interview Survey. A rotating panel. About 13,000 addresses are contacted each quarter, and usable interviews come from about 5,000 households each quarter. A consumer unit stays in the sample for four consecutive quarters and is interviewed every three months.
- Diary Survey. About 18,000 addresses a year are selected, and usable diaries (two one-week diaries per household) come from roughly 6,700 households.
- Population. The civilian noninstitutional population, including group quarters such as college dormitories. Military personnel overseas or on base, nursing home residents and prison inmates are excluded.
The Handbook's data sources page says the Interview Survey collects expenditures "usually fairly easy to recall for periods of 3 months or longer." Vehicle insurance is a regular, recurring item, and it is listed under transportation in the published tables. The tables guide adds that diary data cover one week and are multiplied by 13, and that the published tables combine the two surveys.
Two points that readers often miss:
- Averages include non-buyers. The guide states that "averages are calculated using all CUs rather than only those that purchased the item." A consumer unit with no car, or one that goes uninsured, counts as zero. The $2,153 is therefore not "what a car-owning household pays."
- The sample is small relative to the country. Roughly 5,000 interviews per quarter represent about 138 million consumer units. BLS publishes standard errors and flags estimates with a relative standard error of 25% or more as unreliable. The estimates discussed below all fall well under that line.
The 2025 numbers by income
The CE's annual table of income quintiles splits consumer units into five equal groups by income before taxes. The 2025 vehicle insurance line, with the income and vehicle counts reported in the same table (the last two columns are our arithmetic):
| Income group | Mean income before taxes | Vehicle insurance, mean per year | Vehicles per consumer unit | Insurance as % of income | Insurance per vehicle |
|---|---|---|---|---|---|
| Lowest 20% | $18,125 | $1,011 | 1.1 | 5.6% | about $920 |
| Second 20% | $46,080 | $1,758 | 1.6 | 3.8% | about $1,100 |
| Middle 20% | $78,584 | $2,199 | 1.9 | 2.8% | about $1,160 |
| Fourth 20% | $126,349 | $2,661 | 2.4 | 2.1% | about $1,110 |
| Highest 20% | $286,188 | $3,147 | 2.7 | 1.1% | about $1,170 |
| All consumer units | $110,847 | $2,153 | 1.9 | 1.9% | about $1,130 |
Two readings are worth drawing out.
Dollars rise with income; the burden falls. Dollar spending climbs steadily across the groups, but the share of income drops from 5.6% to 1.1%. A fixed-price product weighs more on a small income, which is the intuition behind affordability work such as The Federal Insurance Office's study of auto insurance affordability.
Most of the dollar gap is vehicles, not price. Dividing by the vehicle count suggests spending per vehicle is fairly flat from the second group up, between roughly $1,100 and $1,170, and lower in the bottom group. Higher-income units own more vehicles, so they pay more in total. This per-vehicle figure is rough: vehicle counts are rounded to a tenth, include vehicles that may not carry insurance, and the CE does not say which coverages were bought. It is a way to see the pattern, not a rate.
The BLS table's own "share" column is different: it shows insurance as a share of total spending. In 2025 that was 2.6% for all consumer units, 2.7% in the lowest group, 3.4% in the second, 3.2% in the middle, 2.9% in the fourth and 2.0% in the highest. Spending shares and income shares answer different questions, so check which one a story is quoting.
By age of the reference person
The CE classifies each consumer unit by the age of its reference person. The age table for 2025 shows mean annual vehicle insurance spending of:
| Age of reference person | Mean vehicle insurance spending |
|---|---|
| Under 25 | $1,443 |
| 25 to 34 | $1,995 |
| 35 to 44 | $2,283 |
| 45 to 54 | $2,816 |
| 55 to 64 | $2,406 |
| 65 and older | $1,734 |
| 65 to 74 | $1,964 |
| 75 and older | $1,434 |
The pattern is an arch: low for the youngest units, highest at 45 to 54, lower afterward. It does not resemble the premium-by-age curve that insurers use, where teenagers and young adults are the costliest drivers. That mismatch is the point. The CE counts a household's total bill, so the number of cars and drivers on the policy matters more than the age rating of one driver. A likely explanation, which the survey itself does not test, is that households in their 40s and 50s tend to have more vehicles and more drivers on the policy, while the youngest and oldest units tend to have fewer. For the rating side of age, see rating factors and the guide how car insurance rates are calculated.
By region
The region table for 2025 reports:
| Region | Mean vehicle insurance spending | Share of total spending |
|---|---|---|
| Northeast | $1,913 | 2.1% |
| Midwest | $1,649 | 2.0% |
| South | $2,400 | 3.4% |
| West | $2,379 | 2.6% |
The South stands out on share because it combines high insurance spending with lower average incomes: BLS shows mean income before taxes of $98,543 in the South against $129,961 in the Northeast. The regions are broad, however, and each contains states with very different premiums and very different laws. A reader in Michigan or Florida cannot assume the regional figure describes their state. For state-level dollar figures, use the NAIC's state tables explained in What NAIC auto insurance data measures, and for burden by state see car insurance affordability by state: premium as a share of income.
Spending is not price
The most common mistake with this survey is reading spending as price. Three things stand between them.
- Coverage choices. A household that carries state minimum limits only will report less than one that carries liability at high limits, collision and comprehensive. The CE does not record coverage.
- Number of vehicles and drivers. As the table above shows, count of vehicles drives much of the difference across income groups.
- Non-purchasers and the uninsured. Averages include everyone. If more households go without insurance because of cost, the average spending falls even as premiums rise.
Price indexes handle these problems by freezing the product; the CE does the opposite and shows what happens after households adjust. Compare the CPI's measure of price, discussed in The CPI for motor vehicle insurance: how price change is measured, with the CE's measure of outlay, and the differences become a feature rather than a contradiction. The same gap between spending and price is treated at more length in average expenditure vs average premium vs quoted rate.
Why the CE and the NAIC disagree
The NAIC's Auto Insurance Database Report reports a 2023 countrywide average expenditure of $1,281 per insured vehicle, computed from insurers' written premiums divided by written car-years and drawn from statistical agents. A quick comparison with the CE's $2,153 per consumer unit in 2025 seems to leave the CE far lower on a per-vehicle basis: dividing by 1.9 vehicles gives about $1,130. But the two numbers are not built on the same basis:
| NAIC average expenditure | CE vehicle insurance | |
|---|---|---|
| Source | Insurer premium and exposure data from statistical agents | Households' own reports |
| Unit | Insured vehicle (car-year) | Consumer unit |
| Includes uninsured vehicles | No | Yes, in the vehicle count and the denominator |
| Year in the tables discussed | 2023 | 2025 |
| Coverage included | Liability, collision, comprehensive | A single line called vehicle insurance |
The NAIC itself warns that its report reflects what consumers spend on average but not how much insurance they are buying. The CE has a similar limitation, plus an extra: it relies on respondents' memory and willingness to report.
We did not find a statement on the pages we reviewed that defines exactly which policies the CE's "vehicle insurance" line includes, so do not assume it is limited to personal auto. And because the NAIC figure is two years older, a straight comparison also mixes in the price increases of 2024 and 2025 shown in the CPI.
Limits of the survey
Several limitations deserve attention before you quote a number.
Recall and reporting error. The Interview Survey asks about spending over the prior three months, and insurance can be billed monthly, semiannually, annually, or inside a bundle with a home policy. Respondents may not remember or may misallocate an amount. BLS counters part of this with a re-interview program and a computer-assisted interview that shows expenses reported earlier to prevent duplication.
Underreporting. A National Academies panel that reviewed the CE concluded: "Underreporting of expenditures is a major quality problem with the current CE." The panel found underreporting differs sharply across kinds of purchases and was most pronounced for small, irregular items; larger, regular purchases were reported better. It compared the CE with the national accounts' personal consumption expenditures for comparable categories and found the ratio falling from 0.88 in 1992 and 1997 to 0.81 by 2007, while cautioning against treating the national accounts as truth. Vehicle insurance is a larger, regular purchase, which is the category the panel found better reported, but the panel's finding is general and we did not find a vehicle-insurance-specific estimate.
Who answers. Some households refuse to take part, and the same panel noted that a growing number of higher-income households live in controlled-access residences, which makes contact harder. BLS adjusts weights for nonresponse using region, consumer unit size, contact attempts and IRS ZIP-code income data, and calibrates to known population counts. Even so, spending at the very top of the income range is the hardest to capture, so treat the highest-quintile figure with the most caution.
Sampling error. Every figure is an estimate with a standard error. The 2025 vehicle insurance mean for all consumer units has a relative standard error of 1.56%, but the under-25 group is 8.23% and the lowest-income group 4.85%. Differences of a few percent between adjacent groups may not be real.
Annual snapshots. The tables are calendar-year averages, released about ten months after year-end, and are not designed to show monthly change.
Using the tables yourself
A short procedure for checking a claim about household insurance spending:
- Go to the BLS CE tables page and choose the calendar-year table you need: income quintiles, age, region or another breakdown.
- Find the row labeled Vehicle insurance under Transportation. Read the Mean line for dollars, Share for the percentage of total spending, and the RSE line for reliability.
- Note the year in the table title and the number of consumer units. Spending is per consumer unit, not per person or per vehicle.
- If you need burden, divide by the mean income before taxes from the same table. The CE does not publish that ratio.
- Before comparing with another source, list differences in unit, year, coverage and who is counted. The comparison in the previous section is the template.
For a household of your own, a reality check is easier: add up the premiums on your declarations page and divide by your pre-tax income. If you are well above the average for your income group, the car insurance quote comparison worksheet can help you test whether your coverage and price are in line.
Common questions
How much of a household budget goes to car insurance?
On the 2025 CE figures, vehicle insurance was 2.6% of average annual expenditures for all consumer units, or about 1.9% of average income before taxes by our calculation. The share varies widely: roughly 5.6% of income in the lowest quintile and 1.1% in the highest.
Does the CE tell me what I should pay?
No. It reports averages over all kinds of households, including those with no vehicle, with different cars, drivers and coverages. Use it to judge budgets and trends. To judge price, get quotes for your own profile.
Why does the FIO study talk about 2%?
The Treasury's Federal Insurance Office chose two percent of household income as its affordability benchmark partly because CE data showed households spending about that share. The study's measure used liability premiums rather than total household spending; see The Federal Insurance Office's study of auto insurance affordability.
How often are the numbers updated?
The CE publishes annual tables once a year. The 2025 tables were released in October 2026 according to the BLS source line. Earlier years remain available for comparison, and BLS does not routinely revise published data.
Key takeaways
- The 2025 CE shows average vehicle insurance spending of $2,153 per consumer unit, about 1.9% of pre-tax income, with a burden of roughly 5.6% in the lowest income fifth and 1.1% in the highest.
- Spending rises with income mostly because higher-income units own more vehicles; spending per vehicle is much flatter.
- CE averages count households with no vehicle or no policy as zero, so they are not what a car owner pays and do not match NAIC per-vehicle averages.
- The survey relies on self-reports; a National Academies panel called underreporting a major quality problem, and the estimates carry sampling error that grows in small groups.
- Use the CE for budget and burden questions, the CPI for price change, and the NAIC for insurer-reported premium averages, and always check the year and the unit.