What NHTSA counts in the economic cost of motor vehicle crashes

The National Highway Traffic Safety Administration estimates that motor vehicle crashes in 2019 cost the United States $340 billion in economic terms, about $1,035 for every resident, and about $1.37 trillion once the report puts a dollar value on lost quality of life. Private insurers paid roughly 54% of the economic total, so crash costs reach your premium, but only the part insurers actually pay is priced into it; congestion, public spending and pain and suffering are not.

The report is the standard citation for "what crashes cost America." This article explains what it counts, how it builds the totals, who pays each piece and where its numbers should not be carried into a conversation about your own insurance bill.

The report in brief

The study is The Economic and Societal Impact of Motor Vehicle Crashes, 2019 (Revised), NHTSA report DOT HS 813 403. Its authors are Lawrence Blincoe, Ted Miller and colleagues from NHTSA, the Pacific Institute for Research and Evaluation and the Virginia Tech Transportation Institute. It was first published in December 2022; a February 2023 revision added an appendix on quality-adjusted life years and corrected some tables, with, in NHTSA's words, no conclusions or findings affected.

Its headline numbers for 2019:

• 36,500 people killed, 4.5 million injured and about 23 million vehicles damaged, including police-reported and unreported crashes.
• Economic cost of $339.8 billion in 2019 dollars, equal to 1.6% of U.S. gross domestic product.
• Societal harm, which adds quality-of-life valuations, of $1.37 trillion.

"Motor vehicle crashes" here means crashes on roadways. The report's footnote says it does not include off-road or parking-lot crashes.

What goes into the $340 billion

NHTSA calls these "human capital" costs: the resources used, or needed, to restore crash victims as far as possible to their pre-crash physical and financial status, plus property and system costs. The main components, from the report's summary tables and executive summary:

The first five rows come from the executive summary; the last four are read from Table 1-1 of the report and are rounded here, so use the report itself for exact figures. Two points stand out. Property damage is the largest single piece, and crashes in which nobody was hurt (the report calls them property-damage-only crashes) account for $101.3 billion, nearly 30% of the total. And lost productivity, at $106 billion between work and household, is larger than medical costs, because a serious injury or death removes years of earnings and unpaid work such as childcare and housework.

For individual cases, the report puts the average lifetime economic cost of a fatality at $1.6 million, with over 90% of that from lost workplace and household productivity and legal costs. Each critically injured survivor, defined by an "MAIS5" score on the Abbreviated Injury Scale, costs an average of $979,000, with medical costs and lost productivity making up 81% of that.

How the report builds the totals: the unit-cost method

The report's method is a multiplication. For each injury severity level it estimates a unit cost per person: medical care, emergency services, lost work, lost household production, legal and insurance costs, congestion and property damage. Then it multiplies by the estimated number of people at that severity.

Counting. Chapter 2 builds incidence: police-reported deaths and injuries, plus estimates for unreported crashes. The report says about 60% of property-damage-only crashes and 32% of all injury crashes were not reported to police, and that unreported crashes account for 19% of economic costs. Severity is classified on the Abbreviated Injury Scale (a six-point system from minor to maximal), with translators from the KABCO scale that police reports use.

Costing. Medical and work-loss costs come from hospital and emergency department discharge files and death records. Property damage and insurance-related costs come from insurance data. NHTSA bought 2018 data from the Insurance Services Office, a pooling organization, which covered 26.3% of private passenger auto premiums, and used it to estimate property damage per vehicle, the number of people insurers compensate for injury, and claim-handling and legal costs.

Present value. Costs of injuries occurring in 2019 are the present value of lifetime costs, expressed in 2019 dollars at a 3% discount rate. A death at age 30 therefore carries decades of lost earnings, discounted to 2019.

Scaling. The result is a total for all crashes and, in later chapters, sub-totals for alcohol-involved, speed-related, distraction-related, motorcycle, pedestrian, rural and urban crashes. Alcohol-involved crashes were $68.9 billion (20%) of economic costs and distraction-related crashes $98.2 billion (about 29%). Those categories overlap with each other and are not additive.

Who pays: the insurance share

NHTSA also traces who ends up paying each cost. Its executive summary gives these shares of the total:

The public share cost taxpayers about $30 billion in 2019, equal to $230 in added taxes per household. Overall, NHTSA concludes, those not directly involved in crashes pay roughly three-quarters of all crash costs, mostly through insurance premiums, taxes and congestion costs, a total over $261 billion.

That is the connection to your premium: premiums are the main channel through which strangers share one another's crash costs. If you want the mechanics of how an insurer turns claim costs into a price, see how car insurance rates are calculated and claim frequency vs severity: which one actually moves premiums?.

What the insurance chapter adds

Chapter 3 contains pieces that are useful on their own, all from 2018 insurance data:

• Insurance compensated about $83.1 billion in crash property damage in 2018.
• Insurers paid for damage to an estimated 18.3 to 18.8 million crashed vehicles, and, apart from uninsured motorist coverage, 4.62 to 5.14 million injury claims.
• Claim administration adds 16.46% to loss costs: 4.27% for defense and cost containment and 12.19% for adjusting, according to ISO and Insurance Information Institute data cited in the report.
• NHTSA estimated that auto insurance compensated about 39% of crash injury costs, including 76% of medical costs and 35% of work losses.
• It estimated that about 9.7% of the $73.8 billion paid to compensate bodily injury went to fraudulent or built-up claims, noting that a 2015 Insurance Research Council study had estimated fraud and build-up at $6.2 to $8.6 billion for 2012 in 2019 dollars. See insurance fraud for how that term is used.
• It cited the Insurance Information Institute (2022) for an estimated 12.6% of U.S. drivers uninsured in 2018, a reminder of why uninsured motorist coverage exists. The Insurance Research Council series on the Triple-I's uninsured motorists page now lists 12.1% for 2018, so estimates for the same year differ by edition.

Only 39% of injury costs being compensated by insurance explains something readers often miss: the people hurt in crashes absorb a lot of the cost themselves, through lost income, out-of-pocket expense and pain, which is why the report notes that "the most serious injuries are not adequately covered by insurance."

Economic cost versus societal harm

The report presents two measures and cautions against mixing them up.

Economic cost counts resources: money spent or income lost. It can be measured from bills, wage data and claims files.

Societal harm (the report also calls it comprehensive cost) adds a dollar value for lost quality of life, meaning pain, disability and the years of life lost. The report bases it on U.S. Department of Transportation guidance, which set a value of a statistical life of $10.9 million in 2019 dollars. Using that value, plus quality-adjusted-life-year factors for nonfatal injuries, the report reaches $1.37 trillion, nearly four times the economic total. Of that, 75% is lost quality of life, and the lifetime comprehensive cost of a fatality is $11.3 million, 86% of it quality of life.

The report is candid about uncertainty: it says the quality-of-life estimates are less certain than the economic ones, and that published estimates of the value of a statistical life range from a few million dollars to over $30 million. Using the DOT's suggested sensitivity range of plus or minus 40% on the value of a statistical life, its Appendix A gives a range of $869 billion to $1.76 trillion for societal harm. It also says economic costs should not, by themselves, be used to produce cost-benefit ratios for safety measures, because they omit the intangible consequences.

The practical lesson for an insurance reader: if you see "$1.4 trillion" in a headline, that figure includes a value that no insurer and no claims file pays. The $340 billion economic figure is the one that reflects money changing hands, and even it includes categories that insurers do not pay.

A worked example: counting one crash

These numbers are invented to show which category each dollar falls into and who might pay it. They are not averages.

Say a driver is rear-ended by a distracted driver at an intersection. The injured driver's costs:

• Repairs to the car: $6,000 (property damage).
• Emergency room and follow-up care: $9,000 (medical).
• Three weeks off work: $4,500 in lost wages (market productivity), plus the value of household tasks someone else must do (household productivity).
• Ambulance and police response (emergency services).
• Forty other drivers delayed by an average of ten minutes while the lane is blocked (congestion).
• A liability claim handled by the at-fault driver's insurer, with an adjuster and some legal work (insurance administration and legal costs).

If the at-fault driver is insured with sufficient limits, that driver's property damage liability coverage pays the $6,000 and bodily injury liability coverage pays the medical bills and wage loss, $13,500 in this example, so insurers pay $19,500 in losses. Applying NHTSA's national 16.46% claim administration load, administrative cost on those losses would be about $3,210. That load is a national average from the report, not any insurer's actual expense ratio.

In NHTSA's accounting, the economic cost of the crash is all of the above plus congestion and public services. The victim bears the pieces the insurer does not pay, such as a deductible if they used their own collision coverage, or wages beyond what any policy replaces. Delayed drivers pay in time and fuel. In your premium, only the $19,500 of paid losses and the insurer's expenses ever appear, and they appear as a small contribution to a pool of many claims, not as a bill to you.

Why NHTSA's totals are not a premium formula

Several features of the report make it a poor input for pricing, and a poor basis for claims such as "crashes cost each driver X in insurance."

1. It counts everyone's costs. Congestion, public services, employer costs and lost household production are real costs but are not claims an insurer pays. The report's own table shows that 46% of economic cost falls on someone other than a private insurer.
2. It is national and averaged. Rates are filed and regulated state by state and built from state claim experience by coverage. NHTSA's report does include state costs in its own chapter, but a state average of all crash costs is not a state's insured loss cost. The report itself says average costs for injuries can be exceeded in individual cases by a factor of three or more.
3. It is one year, in 2019 dollars. Premiums respond to current repair costs, medical prices and claim frequency. The Triple-I's auto facts page, which reprints Bureau of Labor Statistics data, shows how far prices have moved since: from 2019 to 2025 the motor vehicle insurance index rose from 571.0 to 893.5 (about 56%), motor vehicle body work from 313.5 to 432.8 (about 38%), and all items from 255.7 to 321.9 (about 26%). A 2019 cost study cannot tell you today's repair bill.
4. It is not the same as insurers' incurred losses. The same Triple-I page, citing NAIC data, lists total incurred losses for private passenger and commercial auto of $196.0 billion in 2020 and $261.4 billion in 2024. Those are claims on policies in a year, a different quantity from NHTSA's present-value cost of crashes in 2019, so the two should not be compared as if one were a mistake.
5. Fault and coverage matter to who pays. In a no-fault state, your own policy pays part of your medical costs; in a tort state, the at-fault driver's liability coverage pays. The report's cost totals do not tell you which coverage bore a given dollar. See Do no-fault states really cost more? Comparing premiums by system.
6. Prevention math is different from pricing math. NHTSA built the study to evaluate safety programs, such as seat belts, which it says prevented 14,653 fatalities and saved $93 billion in economic costs in 2019. A cost-benefit study of a seatbelt law asks what society saves; a rate filing asks what the insurer is likely to pay.

For how crash data does enter pricing, see Crash statistics and premiums: reading NHTSA FARS and IIHS data and how medical costs and injury claims shape what you pay for liability.

How often the figures are updated

NHTSA does not publish this study every year. The 2019 report says NHTSA last examined crash costs in 2015, in a report based on 2010 data, which estimated $242 billion. The 2019 estimate is about 40% higher. NHTSA attributes the increase to several factors: inflation of about 17%, changes in injury severity and outcomes, a revised basis for unit costs, more deaths and injuries from added driving, and property damage costs that rose as the fleet shifted toward more expensive light trucks and SUVs and as newer vehicles carry more safety equipment that must be replaced after crashes. The value of a statistical life also rose 23% over the 2010 basis.

As of November 2026, the 2019-data report is the most recent edition I found; I did not find a newer NHTSA cost study. If you cite it, give the data year (2019), the publication date (December 2022, revised February 2023) and the dollar year (2019), and check NHTSA's publications list for a newer edition. Updating the dollar amounts by a price index would be your own calculation, not NHTSA's, and should be labeled that way.

Common questions

Is $340 billion what insurers pay out in a year?

No. It is the present value of lifetime economic costs of crashes that occurred in 2019, borne by victims, insurers, governments and others. Insurers' share was about 54%, and the claims they paid in a year are measured separately in NAIC data, discussed in Insurance Information Institute statistics and how to use them.

How does this connect to why my premium goes up?

Indirectly. Crash costs that insurers pay are the main input to rates, so rising repair, medical and injury costs push rates up. But NHTSA's report describes one year, 2019, before the price increases of 2021 to 2024. For the recent drivers of premium changes, see why car insurance got more expensive, and where prices stand in 2026.

Does NHTSA's figure include crashes where nobody was hurt?

Yes. Property-damage-only crashes account for $101.3 billion, nearly 30% of the economic total, because they are so numerous: the report says about 19.3 million of the 22.9 million damaged vehicles were in crashes with property damage only.

Why is the societal number so much bigger?

Because it prices quality of life using a value of a statistical life, $10.9 million in 2019 dollars under DOT guidance. Three-quarters of the $1.37 trillion is lost quality of life. It is a policy valuation, not money spent.

Key takeaways

• NHTSA's 2019-data report puts economic crash costs at $340 billion and societal harm at $1.37 trillion; the second includes a valuation of lost quality of life.
• The economic total is built by multiplying unit costs per injury severity by estimated incidence, including crashes that never reach police reports, in present-value 2019 dollars.
• Private insurers paid roughly 54% of economic costs; individuals paid about 23%, third parties about 14% and public revenues about 9%.
• The report is a national cost-of-crashes study, not a rating tool. Cite it with its data year and do not equate it with insurers' claim payments.
• The 2019 edition is the latest I found as of November 2026, and the dollar figures predate the rise in repair and insurance prices that followed.