NHTSA Just Spent $665 Million Fighting 20% of the Problem

According to the toxicology reports, and there are roughly 36,640 fatal crashes' worth of them from 2025 alone, impairment is a factor in about one out of every five drivers who dies on American roads, not one in two, not one in three, but one in five.^[1] NHTSA just allocated $665 million in state highway safety grants, and the overwhelming majority funds DUI checkpoints, seatbelt enforcement campaigns, speed traps, and distracted driving blitzes.^[2]

That is a lot of money pointed at the smaller slice of the pie.

I pulled every toxicology record in our FARS dataset spanning 2014 through 2023. By vehicle class: sports cars lead at 22.5% impaired, sedans sit at 20.4%, pickups at 20.1%, SUVs at 19.5%, and vans bring up the rear at 18.1%.^[1] Remarkably consistent across segments, which tells you that impaired driving is not a vehicle-type story but a fixed-percentage human behavior story, accounting for roughly a fifth of the body count regardless of whether the driver was behind the wheel of a Mustang or a Sienna.

Now flip the ledger: eighty percent of drivers in fatal crashes were sober, stone cold sober with no BAC and no drug-positive tox screen, and they still died because old engineering killed them.

Model year analysis in FARS reveals something that should embarrass every enforcement-first budget: the Ford Ranger's fatality count dropped 92.6% from model years 2000-2008 to model years 2016-2022, plummeting from 172.6 average annual deaths to 12.8. The Chevrolet Tahoe fell 91.5%, and the Ford Expedition 90.9%, with the fifteen most-improved models averaging an 84% reduction.^[1] No checkpoint achieved anything remotely close to that, no billboard campaign came within an order of magnitude, and no "Click It or Ticket" push even registered on the same scale. Structural engineering, side curtain airbags, electronic stability control, and better crumple zone geometry did what $665 million in annual enforcement grants have been trying to do for decades, and did it at a rate that makes the comparison almost cruel.

And the vehicles still killing people at the highest rates are ancient, relics of an era that should have ended a decade ago. The Chevrolet GMT-400 pickup averages 23.7 years old at the time of fatal crash, the S-10 averages 21, and the GMC Sonoma sits at 20.2.^[3] These are vehicles designed when the federal frontal crash test speed was 30 mph and side-impact standards were essentially optional. Every year one stays on the road, it carries a fatality risk that a 2020 model of the same class simply does not.

So what would $665 million buy if you aimed it at the 80% instead of the 20%?

Run it as a targeted cash-for-clunkers program focused on the ten deadliest vehicles in FARS. At $5,000 per trade-in, that is 133,000 vehicles retired. If model year improvements deliver even half of their observed 84% average fatality reduction per vehicle, you are purchasing a larger absolute reduction in deaths than enforcement achieves, because enforcement's best-case impact on impairment fatalities is 7-9% according to RAND Corporation meta-analyses of DUI checkpoint programs.^[4] Apply that 7-9% to the ~7,250 impairment-addressable deaths in 2025 and you get 520 to 650 lives saved. At $665 million, that works out to $1.02 million to $1.28 million per life. The DOT's own value of a statistical life is $13.2 million, so the investment clears the cost-benefit bar, but barely, and only if you assume every checkpoint dollar translates perfectly into deterrence, which anyone who has ever watched a driver U-turn before a sobriety checkpoint knows it does not.

Strongest Counterargument

Enforcement has a behavioral multiplier that fleet turnover cannot match. A visible DUI checkpoint changes the calculus for every driver who passes it, not just the one who gets arrested. Deterrence radiates. And the original Cash for Clunkers in 2009, which spent $3 billion without targeting the deadliest vehicles specifically, produced mixed safety outcomes precisely because the trade-in criteria were based on fuel economy, not crash survivability.^[5] A safety-targeted version has never been tried at scale, which means the 84% improvement figure is a projection, not a measured policy outcome. Fair. But projections derived from nine years of FARS fatality data across every major model are not speculation. They are the closest thing to a controlled experiment that American roads have ever run, with every model year cohort serving as its own treatment group.

Methodology

Impairment percentages calculated from FARS toxicology records, 2014-2023, using any_pct (any impairment detected: BAC > 0 or drug-positive). Model year improvements use average annual deaths for model years 2000-2008 versus 2016-2022 within each make/model. The $1.02-1.28M per life figure divides the $665M grant total by estimated lives saved, applying the RAND 7-9% DUI checkpoint effectiveness range to the impairment-addressable subset of 2025 preliminary fatalities (36,640 × 0.20 = 7,328, rounded to ~7,250). Fleet retirement estimates assume $5,000 per vehicle; no administrative overhead is modeled, which would reduce the effective count.

Limitations

FARS captures fatal crashes only. The $665 million in grants also addresses injury crashes, which outnumber fatalities roughly 180 to 1, and enforcement's proportional impact on injury severity may differ substantially from its effect on fatality counts. FARS toxicology completion is inconsistent: not all jurisdictions test all fatal crash drivers, meaning the 20% impairment rate likely undercounts. Model year fatality improvements conflate engineering advances with fleet age effects, survivor bias, and changes in who buys which vehicles over time. Grant funds are allocated to states with discretion over spending; the federal priority categories do not perfectly predict how money is deployed locally. Zero dollars of this analysis accounts for the political economy of traffic safety, where enforcement funding has bipartisan support and vehicle retirement programs do not, which is itself a data point about why the allocation persists despite the math.

What this means for you: If you drive a vehicle with a model year before 2010, your car is the intervention nobody is funding. Check your vehicle's NHTSA safety rating at nhtsa.gov/recalls and look at your model's crash test performance relative to current standards. A 2008 five-star rating and a 2022 five-star rating are not the same test and not the same protection. If you are shopping for a replacement, prioritize vehicles with autonomous emergency braking, lane departure warning, and electronic stability control as standard, all of which post-2018 models overwhelmingly include. You cannot lobby Congress to reallocate $665 million, but you can retire your own GMT-400.