NHTSA Just Spent $20 Million Chasing 20% of the Problem

On Sunday, NHTSA Administrator Jonathan Morrison announced a $20 million grant program to fund "cutting-edge traffic safety enforcement innovations."^[1] Predictive analytics for deployment, high-visibility checkpoints, specialized DUI training, state-of-the-art safety tech to "curb dangerous driving behaviors." It sounds great, it photographs well, and it addresses roughly one-fifth of the problem.

We ran toxicology results across every driver in our FARS dataset spanning 2014 through 2023: 490,736 people behind the wheel in a fatal crash who had toxicology data recorded.^[2] Of those, 392,388 had zero alcohol and zero drugs in their system. Eighty percent. Four out of five drivers involved in a fatal crash were stone-cold sober, fully alert, and presumably capable of passing any checkpoint that NHTSA's new enforcement grant could fund.

The remaining 20% breaks down like you'd expect: 15.1% tested positive for alcohol, 8.7% for drugs, with some overlap between the two categories pulling the combined "any impairment" number to 20.0%. The split is remarkably uniform across vehicle classes, ranging from 18.1% for vans to 19.5% for SUVs, 20.1% for pickups, and 20.4% for sedans, with sports cars running slightly hot at 22.5% because of course they do. But that four-point spread across every class of vehicle on American roads is a rounding error compared to the gulf between what enforcement can catch and what it fundamentally cannot.

The ITSE grant doesn't only target impairment, to be fair. It also covers distracted driving, seatbelt enforcement, speeding, and what NHTSA vaguely calls "reckless driving." Those are real problems that kill sober drivers. But NHTSA's own data tells us something uncomfortable about the scale: the agency estimates about 36,000 Americans die in traffic annually.^[3] That $20 million divides out to roughly $556 per death. For context, ESC was mandated in 2012 at a cost of about $100 to $300 per vehicle, and the IIHS estimates it prevents 56% of single-vehicle fatal crashes for every car that has it, saving an estimated 2,200 lives a year and counting.^[4] One engineering mandate, baked into every new car sold, outperforming decades of enforcement grants without a single traffic stop.

This is what makes the $20 million announcement so cosmically frustrating. Enforcement is not worthless: DUI checkpoints save lives, and seatbelt campaigns have measurably moved the needle over decades. Speed enforcement works in specific corridors for specific stretches of time until the officers leave and speeds drift back up. But enforcement is inherently temporary, inherently local, and inherently limited to the behaviors it can observe. A corroded micro-switch in a Mercedes door lock that prevents the parking brake from engaging isn't going to show up on a predictive analytics dashboard, and neither is a timing belt in a Ford EcoBoost that degrades and seizes the engine at 70,000 miles with no warning. Those are engineering problems that killed sober, attentive, law-abiding drivers, and enforcement never had a chance at stopping any of them.

Meanwhile, NHTSA spent this past week on other things too: investigating 1.2 million Teslas over suspension failures^[5], declaring Ford's 1.0L EcoBoost an "unreasonable risk"^[6], and processing Mercedes' 310,667-vehicle door lock recall.^[7] Those three actions alone cover 1.65 million vehicles with engineering defects that could kill their occupants regardless of blood alcohol content, seatbelt usage, or proximity to a funded checkpoint. Zero of those vehicles would have been saved by a $20 million enforcement program, no matter how much predictive analytics you throw at the problem.