The CPI for motor vehicle insurance: how price change is measured

When a news story says car insurance prices rose a certain percent over the past year, the number almost always comes from the motor vehicle insurance index in the Consumer Price Index, which the U.S. Bureau of Labor Statistics (BLS) publishes every month. The index follows the premium for a fixed set of policies whose coverage, vehicle and driver profile are held constant, so it measures how much the same product cost, not how much households spent. For September 2026, BLS reported the index down 5.0% from a year earlier, after sharp increases from 2021 through 2024.

That distinction explains why the headline figure can differ from your renewal notice, from a quote-based study, and from the average spending numbers the NAIC publishes. Each answers a different question. This article walks through how the index is built, what it deliberately ignores, how to read the table on the BLS site, and how to use it without over-reading it.

What the index is, and where it sits in the CPI

The motor vehicle insurance index is a component of the CPI's private transportation group. BLS describes it in its motor vehicle insurance fact sheet as covering physical damage, liability and miscellaneous coverage on private passenger vehicles: collision, comprehensive, bodily injury and property damage liability, medical payments, uninsured motorist and personal injury protection. Cars, vans, SUVs and trucks eligible for private passenger insurance are included.

Three details of scope matter for anyone quoting it:

• It is a national number. BLS publishes the index at the U.S. city average level. There is no official motor vehicle insurance index for your metro area or state, so a local claim that "rates in our city rose 18% according to the CPI" is not something this series can support.
• It is a share of the CPI, not a dollar amount. The index has a "relative importance," its weight in the overall CPI. BLS lists motor vehicle insurance at 2.754% of the CPI as of December 2025, and the September 2026 detailed table shows 2.531 for August 2026. The weight shifts as prices move relative to other goods.
• It is an index number, not a premium. The series (identified as CUUR0000SETE for the unadjusted version) is a level such as 846.1 for September 2026. Only the percentage change between two levels means anything. The level itself is not dollars.

The CPI as a whole is described in BLS's Handbook of Methods as "fundamentally a measure of price change," built around the consumer's out-of-pocket price. The expenditure weights that decide how much each item counts come from the Consumer Expenditure Survey and, per the Handbook, are updated annually. That survey is the subject of a separate article, What households spend on car insurance: Consumer Expenditure Survey.

How BLS follows policies month to month

Most CPI items are priced by data collectors visiting stores or pulling prices online. Insurance cannot be handled that way, because a premium depends on a driver profile, so BLS works with the insurers themselves. According to the fact sheet:

1. Carriers are chosen by size. They are drawn from a universe supplied by a national insurance trade organization, in proportion to the dollar amount of eligible policies each company wrote in a state.
2. Actual policies are selected. BLS works with each cooperating insurer to pick real policies to follow. For each one it records the vehicle's age and description, the garaging location, annual miles, vehicle use, the coverages and limits, and the driver's age, sex, driving record and marital status, plus surcharges, discounts and fees.
3. The profile is frozen. BLS holds those characteristics constant and tracks the premium for that unchanging profile. In BLS's words, it uses "premiums calculated for the policies with these unchanging characteristics and not the actual premiums paid."
4. Prices are collected monthly, spread across three periods within each collection month.
5. The vehicle ages on a schedule. Each year, usually in October or November, each sample vehicle's model year is advanced by one year so that its age stays the same. If the premium changes because of that move, the change is recorded as a price change, up or down.

Two further rules shape the results. Dividends that reduce a premium are recorded as price decreases, while separate cash payments to policyholders are not, because BLS treats those as income. And the sample is old by CPI standards: BLS says the insurance sample was last fully replaced in 2004 and partly replaced in 2018 to add new geographic areas, whereas most CPI components rotate one-eighth of their sample every half year.

What a "frozen profile" does and does not capture

Because the profile is fixed, the index captures rate changes: the insurer raising the base rate, changing a rating factor, or revising a discount. It captures the effect of an insurer re-rating the same 2018 sedan with the same driver at the same address. It does not capture the choices real households make in response.

The fact sheet does not list claims history or accident frequency as separate inputs. Risk characteristics enter only as the fixed profile traits, such as the driver's record, which are held constant. So a driver who has an at-fault accident and sees a surcharge does not pull the index up; that driver's experience is a change in the profile, not a change in the price of the product.

What the index leaves out

The fact sheet lists exclusions, and the design implies others. Reading them together:

None of this makes the index wrong. It makes it a precise answer to a narrow question: how did the price of one unchanging bundle of coverage move?

Price versus bill: a worked example

Say two neighbors each paid $1,500 a year for the same policy in the base year, and the insurer's rate for that bundle rises 20% over the next year. This is a made-up illustration, not market data.

• Neighbor A keeps the same coverage. Her bill rises to $1,800, a 20% increase.
• Neighbor B responds by raising his collision deductible and dropping comprehensive on an older car. His bill rises to $1,550, a 3.3% increase.

A pricing index built on the frozen profile records a 20% price increase, because that is what happened to the product both neighbors originally bought. A survey of what households spent would record something between 3.3% and 20% depending on how many neighbors behave like B. The two measures are not in conflict. One tracks the price of an unchanged product, the other tracks what people ended up paying after they adjusted.

The NAIC makes the same point in its Auto Insurance Database Report. It notes that average premium and average expenditure are affected by the choices consumers make about the types and amounts of insurance, whereas the insurance price index "holds the amount of insurance constant to measure price changes in a uniform product." Its table of annual changes shows the gap: from 2019 to 2023, the NAIC's average expenditure rose 19.24%, the all-items CPI rose 19.37%, and the CPI auto insurance component rose 36.56%. (The report's narrative paragraph gives 21.76% for the CPI figure, which appears to be the new-vehicles row of the same table; the 36.56% in the table is the auto insurance row.) The NAIC's explanation is that when price rises, buyers demand less insurance, by dropping coverage or raising deductibles, so spending rises by less than price. For a longer comparison of these measures, see average expenditure vs average premium vs quoted rate.

Reading the table on the BLS site

Each month BLS publishes the CPI news release with a detailed table of components. The September 2026 version is Table 2 of the CPI release. To find the number behind a headline:

1. Open the table and scroll to the transportation group, then private transportation, then the row labeled Motor vehicle insurance.
2. Read the relative importance column first. It tells you how much weight the item carries, which is context for how much it moves the overall CPI.
3. Read the unadjusted percent change over 12 months. This is the figure used in most headlines. For September 2026 it is -5.0%, meaning the index was 5.0% lower than in September 2025.
4. Read the one-month unadjusted change (-0.3% from August to September 2026) for the most recent direction.
5. Read the seasonally adjusted monthly changes for the last three months (for this row, -0.3%, -0.8% and 0.1% for July, August and September 2026) to see whether the trend is steady.

A reader who wants to compute a change over a longer period can use the underlying series. Using BLS's public data for the unadjusted series, the index stood at 780.284 in December 2023 and 868.417 in December 2024. The change is 868.417 divided by 780.284, minus one, or 11.3%. Because it is a ratio of index levels, it does not matter that the levels themselves look large.

Seasonal adjustment, revisions and a missing month

Seasonal adjustment. BLS publishes both unadjusted and seasonally adjusted data. Headlines on the 12-month change use unadjusted data, since a year-over-year comparison already compares like months. Seasonally adjusted figures are used for month-to-month movement. BLS's seasonal adjustment page says the factors are recalculated each year with the release of the January CPI, which may revise seasonally adjusted indexes for the previous five years; older seasonally adjusted indexes are considered final.

Revisions. BLS's questions and answers page states that the unadjusted CPI-U and CPI-W series are final when issued, while seasonally adjusted figures are subject to revision for up to five years after their release. That is why contracts and cost-of-living clauses typically reference unadjusted data. For your purposes: the 12-month figure you read in a headline will not be restated later; a seasonally adjusted monthly figure might.

The October 2025 gap. The BLS data series carries a note that October 2025 data are unavailable because of the 2025 lapse in appropriations. If you chart the index, expect a missing point that autumn, and be careful with any "monthly" calculation that straddles it. Twelve-month comparisons that start or end on other months are unaffected.

What the recent numbers show

Using the same BLS series, the December-to-December changes since the pandemic look like this (our calculation from the published index levels):

The 12-month change peaked at about 22.6% in April 2024, and the index in September 2026 stood roughly 55% above its December 2020 level. Both are our calculations from BLS data as of the September 2026 release, and they will be restated as new months arrive.

Two cautions about reading a decline. First, a year-over-year drop of 5.0% means the index is below its level a year earlier, not that prices are back where they were before the surge; the cumulative rise is still large. Second, the index tracks a frozen product from cooperating insurers. A household that recently had an accident or moved may still see a higher renewal. The renewal rate increase checker and the guide why car insurance got more expensive, and where prices stand in 2026 cover what drives an individual renewal.

Using the index well

The CPI motor vehicle insurance index is best for three jobs:

1. Checking direction and size of market-wide price change. It is monthly, national, and uses the same method every month.
2. Fact-checking a percent-increase headline. Ask whether the story is quoting the index (a price change for a fixed product) or an average premium (a spending figure). See fact-checking car insurance headlines about percent increases.
3. Putting your own renewal in context. If your premium rose 30% while the index rose 10%, the difference is likely your own profile or your insurer's rate action, not the market. Your declarations page shows your current premium by coverage, which lets you compare like with like.

It is a poor tool for answering "what should my premium be?" or "how much does the average household spend?" For those, see how to read car insurance rate studies and average-cost claims and What NAIC auto insurance data measures.

Common questions

Does the index include claims costs?

Not directly. It measures premiums for a frozen profile. Claims costs reach the index only indirectly, because rising claim costs lead insurers to file higher rates, and those higher rates then show up as higher premiums for the sampled policies.

Why does my renewal differ so much from the headline number?

The headline describes a fixed product at the national level. Your renewal reflects your own profile, location, vehicle, claims and any coverage changes you made, as well as your insurer's rate action in your state. Individual results can sit far from the national average in either direction.

Is there a CPI for car insurance in my state or city?

BLS publishes the motor vehicle insurance index at the U.S. city average level, according to its fact sheet. For state-level spending, the NAIC publishes state averages, with the caution that they should not be compared without accounting for coverage and law differences.

Why did the index fall if I am still paying more than before?

The index fell from a high level. A drop of 5.0% in a year can coexist with a premium that is still well above its 2020 level. If your insurer raised rates earlier and has not cut them, your bill does not follow the index down.

Key takeaways

• The CPI motor vehicle insurance index tracks the price of a fixed set of policies, holding coverage, vehicle and driver profile constant; it is not a measure of what households pay.
• It is published nationally, so it cannot speak for a state or city, and it excludes business vehicles, motorcycles and assigned-risk policies from pricing.
• Headlines use the 12-month unadjusted change, which BLS says is final when issued; seasonally adjusted monthly figures can be revised for up to five years.
• Spending measures from the NAIC rose less than the index from 2019 to 2023 because buyers cut coverage or raised deductibles when prices climbed.
• Read the percent change between two index levels, never the level itself, and check whether a story is quoting an index or an average premium.