Cheap To Mandate, Expensive To Own

This is the second of four essays on the American automotive industry.

The first described the effects that financialization has had on the market. It showed how automaker profits transitioned from vehicle sales to loans, and outlined the impact that this trend has had for consumers.

Today’s article examines the impact that our regulations have had on both consumer choice and long-term cost of ownership.

The next essays in the series will cover the global automotive supply chain and the erosion of our right-to-repair.

A modern car is a complicated web of systems. Each contains three components: a sensor that measures things, an actuator that physically alters things, and a computer that reads sensor states and determines which actuators to actuate.

Those computers are called electronic control units (ECUs). Counting their prevalence over time gives us a good indication of how complicated cars have become.

Manfred Broy, a computer scientist at the Technical University of Munich, found that premium cars contained “not less than 70” ECUs in 2006. A current Volvo XC90 carries upwards of 110, and a BMW 7-series can run to 150 or more.

Beyond electronics and computers, mechanical complications are also on the rise.

The EPA publishes the fleet-wide technology content of every new vehicle sold in America, every year:

The table above contains just a few examples of systems that hardly existed a generation ago and are now ubiquitous. Each comes with its own array of sensors, actuators and ECUs. Each adds to the underlying complexity of the vehicle.

Fitting all of it is cheap. The marvel of modern manufacturing means automakers are able to add all of these systems without massively increasing their costs of production.

The problem is maintaining them. Every sensor is a part that fails. Every controller needs accurate information about the state of the car. When it does not get that information it protects itself by shutting down. A single failed sensor puts the car into limp mode, capped at low speed and reduced power, until somebody diagnoses it.

Every system now has multiple expensive points of failure. Every ordinary repair is now complex.

Consumers have no option to opt-out. There is no bare-bones car on the market.

A lot of this complexity is driven by regulations. Many of them are effective. Let’s take a look at one positive example.

Automatic emergency braking uses a forward-facing camera to watch the road ahead. Some systems add a radar sensor. If the car in front stops and the driver does not react, your car brakes itself. NHTSA made it mandatory in a rule finalized in May 2024, with compliance required by September 2029.

It is cheap. NHTSA costed the whole rule at $354 million a year. Of that, $282 million is software applied to every vehicle. Only $72 million is hardware, because most new cars already have a front-facing camera fitted. Across roughly 17 million new vehicles a year, that comes to about $21 a car.

AEB is also effective. The Insurance Institute for Highway Safety found that automatic braking cuts front-to-rear striking crashes by 50 percent. Crashes causing injuries are reduced by 56 percent.

NHTSA puts the cost per equivalent life saved at $550,000 against the $11.6 million the DOT values a life at. By their own accounting, the rule returns roughly $20 of benefit for every dollar it costs.

This regulation is a no-brainer, given the number of lives it saves on the road.

That being said, there are significant hidden costs that aren’t captured in the agency’s analysis. The crash-detecting camera typically sits behind a windshield and looks through it. Its aim needs to be calibrated to within a fraction of a degree. New glass has a marginally different thickness and curvature, so every windshield replacement now requires a complex calibration: a technician, a target board, a level floor and a scan tool.

The Highway Loss Data Institute compared trim levels of the same models, where the system had been optional. Across more than 270,000 glass claims:

“Hondas with front crash prevention were over 12 times more likely to have a glass claim of $1,000 or more. Subarus with front crash prevention were over 6 times more likely.”

AAA had estimators price the same job. In its 2023 report, the average windshield replacement ran $1,439, of which $360 was attributable to the driver-assistance system.

You cannot decline the calibration to save the money. Under the American national standard for auto glass, a shop must calibrate or refuse the job outright.

Americans replace 11 to 15 million windshields a year. Calibration runs $300 to $400. If most of those cars now need it, the recurring national bill is somewhere between $3 billion and $5 billion annually.

NHTSA costed the rule that created the bill at $354 million.

That is a cost-tail that is roughly 10x the mandate. For an emergency braking system that is shown to be effective in saving lives, that gap may be understandable. For many of our regulations, it is not.

NHTSA publishes a cost-benefit analysis for every rule it issues. The key metric is cost per equivalent life saved. It allows us to compare the costs and benefits of one rule against another, and set those against the value the department itself puts on a human life.

Top to bottom, that is a variance of about a hundred to one. The seat belt figure was derived by dividing the per-vehicle cost by the 15,485 lives NHTSA credits belts with saving in 2012. Every other number is the agency's own.

Three of the rules in the table above failed on the government’s own accounting.

Tire pressure monitoring. The regulatory analysis states that net benefits are negative under every compliance option at both discount rates. The rule went in anyway, in 2007. To satisfy it, manufacturers put a pressure sensor and a radio transmitter inside all four wheels. Each has a sealed battery that cannot be replaced on its own. When it dies after five to ten years you buy a whole new sensor, and the dashboard light stays lit until you do. The rule prevents an estimated 120 deaths a year.

Roof crush resistance. In 2009 the agency tightened the standard and extended it to vehicles weighing up to 10,000 pounds. Its own analysis of that extension is blunt:

“Nearly all alternatives covering vehicles from 2,723 and 4,536 kilograms (6,001 and 10,000 pounds) GVWR yield net losses rather than net savings to society.”

For that class of vehicle the agency expected to prevent two deaths a year, at somewhere between $18.8 million and $140 million each.

Backup cameras. The final rule says it outright:

“While the costs of the rule exceed its quantifiable benefits, Executive Orders 12866 and 13563 call upon us to assess the costs and benefits of a rulemaking, including those costs and benefits that are difficult to quantify.”

The cameras prevent thirteen to fifteen deaths a year, at $15.9 to $26.3 million each. In fairness to the agency, it did not choose to implement this rule itself. Congress ordered it in the Cameron Gulbransen Kids Transportation Safety Act of 2007.

We pay for the rule at the bottom of that table exactly the way we pay for the one at the top. Both are fitted to every car, and no version of any car exists without either.

Safety rules are simple. They mandate that certain equipment be installed on new vehicles. Environmental rules are different.

The EPA set each manufacturer a fleet-average carbon dioxide target, measured in grams per mile across everything it sold that year. For model year 2026 the industry-wide target was 161 grams per mile.

A manufacturer that came in above its own target ran a deficit, and it had three model years to erase that deficit or buy credits from a competitor who had spare ones. If a manufacturer failed at both, then its cars were not covered by a certificate of conformity, which exposed it to a civil penalty of $59,114 per vehicle sold.

So grams were money and therefore credits were money, so the manufacturers went hunting for credits.

A credit does not reduce what comes out of the tailpipe. It reduces the emissions number the manufacturer reports. Finding a real gram of efficiency requires difficult engineering: a better combustion chamber, a lighter body, a smarter transmission. That is expensive and slow. Fitting a credit-earning system is cheap and quick. Both count the same against the target.

The EPA published a list of technologies that earned credits automatically. They are called off-cycle credits, because the claimed benefit does not show up on the standard fuel economy test. They were applied after the true emission testing was done.

Some of the list:

Grams of carbon dioxide per mile, credited on every vehicle fitted. A separate schedule covered exterior lighting, running down to 0.08 grams for a more efficient license plate lamp.

Claiming from that list took no work at all. The manufacturer fitted the technology, looked up the published number, and subtracted it from the fleet emission average it reported to the EPA. There was no test to run, no application to file and nothing to approve. The agency's only power of review was that it "may request" supporting data afterwards.

A manufacturer could take up to 15 grams per mile this way, close to a tenth of that 161-gram obligation, without conducting a single test.

The list just outlines the default credits for each system. A manufacturer that wanted more credits could apply, submit a methodology, justify why ordinary testing would not capture the benefit, and wait for the EPA to publish the application for public comment. Most of the large manufacturers did. Ford applied for 1.2 grams per mile for an improved window anti-fogging strategy. Jaguar Land Rover requested 1.6 grams for a 48-volt motor-generator. Mitsubishi requested 1.1 for an air conditioning compressor.

The whole exercise became a hunt for grams, a fraction of a gram at a time.

This is why the credit structure is so ineffective. It creates perverse incentives. When the government mandates something it has to defend the requirement. The industry then has every motivation to attack the numbers. When it offers a credit instead, manufacturers are incentivized to claim every gram on offer, regardless of how effective they truly are at reducing emissions.

Ventilated seats are a good illustration. The list paid 1.0 gram per mile on a car, on the theory that a cooled seat lets you tolerate a warmer cabin, so the air conditioning runs less. Then manufacturers applied for more.

Fiat Chrysler asked for 2.3 grams on cars and 2.9 on trucks, more than double the published value, citing a National Renewable Energy Laboratory study from 2005 and later work by the seat supplier. A separate request from General Motors rested on a trial in which four men sat in two Ford Focuses in Golden, Colorado for twelve days in September 2016 and reported feeling equally comfortable at a cabin temperature 2.6 degrees Celsius higher. That trial was not blinded, and the heat thrown off by the seat fan itself was left out of the model. That is the evidentiary standard we are talking about.

In model year 2023 the industry averaged 8.6 grams per mile of off-cycle credits, and twelve of the fourteen largest manufacturers finished the year emitting more than their standard allowed. They complied on credits.

Stop-start earned 1.5 grams on a car, and 2.5 grams if the system included an electric heater circulation pump. So manufacturers added the pump. On conventional gasoline engines, stop-start went from 0.6 percent of production in model year 2012 to 57.6 percent in 2024. That is what happens when you incentivize complexity.

Interestingly, in February 2026 the EPA repealed the entire greenhouse gas program. Its own press release described the stop-start credit as "nothing more than a regulatory loophole that allowed automakers to claim GHG credits without delivering real-world emission reductions or benefits to human health."

The credits survived though. The same technologies at the same values now count against fuel economy standards instead, through model year 2032. Stop-start still pays 2.5 grams on a car and 4.4 on a truck.

So despite the EPA's own finding that the credit delivered no real-world emission reductions, nothing comes out of the car. Three-quarters of model year 2026 models still ship with stop-start as standard equipment. It stays there for the life of the vehicle, and so does the bill for maintaining it.

CarMD compiles what American shops find when customers drive in with an illuminated check engine light. It logged more than 39 million vehicle failures and recommended repairs in 2025. The most common one in the country was replacing the catalytic converter, at an average of $1,511. Second was the oxygen sensor. Sixth was the evaporative emissions purge valve. Those three come to 19 percent of the list. Nearly a fifth of what an American check engine light reports is a part fitted to control emissions.

The evap system is a great example of the issue. It is a charcoal canister, a purge valve, a vent valve and a pressure sensor. Its entire job is catching fuel vapor so it does not escape while you refuel. Gas stations in smoggy states were separately required to catch the same vapor at the nozzle. In 2012 the EPA waived the gas station requirement, finding the two systems "redundant" and noting that they sometimes interfered with one another. It let states drop the half that belonged to the gas station. The half bolted to your car stayed. It is still throwing codes. A lit check engine light is an automatic fail on an emissions test, so in the states that require one you cannot renew the registration until somebody has fixed it.

The catalytic converter is another such example. Tightening nitrogen oxide limits worldwide raised the amount of rhodium each converter needs, and rhodium peaked above $29,000 an ounce. That turned a bolt-on part hanging under every car into something worth stealing with a battery saw. Theft claims went from 16,660 in 2020 to 64,701 in 2022. State Farm alone paid $115 million on 45,000 claims in 2022, averaging $2,500 each.

The same pattern runs through the rest of the car.

A stop-start system cycles the starter tens of thousands of times more than a conventional one. That of course wears starters out faster, so manufacturers must include an over-engineered and expensive version to compensate. Worse than that though, an ordinary lead-acid battery will not survive so many cycles, so the car needs an absorbed glass mat battery instead, at 40 to 100 percent more. On several brands, including BMW, Mini, Audi and Volkswagen, fitting one requires a scan tool to register the new battery with the car. Skip that step and the charging system keeps treating it as the old, tired battery it replaced, and the new one dies early. All of that for a job that used to take five minutes in any AutoZone parking lot.

A manufacturer could build a car that meets only the mandatory minimum. Nothing stops it. What they can’t do is sell you one with any of the mandatory items taken out.

The law that governs this is 49 U.S.C. § 30122:

“A manufacturer, distributor, dealer, rental company, or motor vehicle repair business may not knowingly make inoperative any part of a device or element of design installed on or in a motor vehicle or motor vehicle equipment in compliance with an applicable motor vehicle safety standard.”

This law binds every party you could possibly transact with, and all of them are forbidden to hand you the car without the mandated safety features.

The statute in no way limits the owner. You may disable equipment on your own car in your own garage. You are permitted to want it gone. You are not permitted to buy it that way.

Slate Auto is attempting to demonstrate where the simplicity floor sits. It is a new American manufacturer whose entire proposition is deleting unnecessary components: no paint, no infotainment screen, no speakers, crank windows, $24,950 before delivery. Its head of design says the interior is "not corrupted with a bunch of screens that are useless to people."

It ships with a screen anyway, behind the steering wheel, because FMVSS 111 requires the rear camera image to appear within two seconds of shifting into reverse.

Two machines, both on sale new in America today.

A Suzuki DR650S costs $7,299. Air-cooled, single cylinder, carburetor, no ABS, a design essentially unchanged since 1996. Its owner can fix it in the field, as I have done many times.

A BMW R 1300 GS starts at $20,395 and can be specced to $30,000+. Four riding modes, cornering ABS, traction control governed by an inertial measurement unit, radar cruise control and an electromechanical clutch.

People ride both on the same trails. One buyer wants the most capable machine available and will pay for it. The other wants something agricultural that they can repair themselves and that will still run in thirty years. The motorcycle market serves both of them. The car market serves only the first.

The regulatory difference is the reason. Cars must meet dozens of federal safety standards. Motorcycles are subject to only a handful of restrictions affecting brakes, lights, mirrors, tires and glazing. None of them govern what happens to a rider in a crash.

The DR650 shows what regulation does to consumer choice across different markets. Sales were halted in Europe around 2001, because a carbureted engine cannot pass European emissions standards. It left Australia in November 2021 when an exemption to that country's anti-lock brake mandate ran out.

There is a fair objection here. A motorcycle crash usually kills only the rider, and motorcyclists die at about 28 times the rate of car occupants per mile. Riding is closer to a risk you take on yourself, so it makes sense to regulate it differently.

But if that principle governed car regulation, we would expect the heaviest rules to fall on the harm a car does to other people.

They do not. Washington has spent sixty years regulating how a car protects the people inside it, and rather less on how it affects the people outside. It has never written a rule about what the front of the car does to a pedestrian it actually strikes. Nothing covers the hood, the general size or the shape of the nose. NHTSA conceded as much when it proposed one in 2024, saying it would add pedestrian crashworthiness requirements to the federal standards "for the first time."

Technology itself is not the problem. Automatic braking prevents half of all front-to-rear collisions for about twenty dollars a car. Stability control saves more than a thousand lives a year. Anti-lock brakes are close to free and they work.

Nor is this an argument against regulating cars. Driving imposes costs on people who never agreed to bear them. That is what regulation is for.

The problem is that rules go in easily and only come out with great difficulty. Tire pressure monitoring went in with negative net benefits and has never been revisited. NHTSA published a finding that the backup camera rule cost more than it saved and issued it anyway because Congress ordered it.

I do not know what the fix looks like. Sunset clauses expire good rules along with the bad. Periodic review could work, but also adds to the already inflated bureaucratic load. If you have seen an efficient policy reform process that works, in any domain, then send me a note. There must be a better way.

Until we figure it out, consumers can expect to pay a steadily increasing amount for repairs that were once routine, or not necessary at all. Windshields, wheel sensors, converters, batteries that need a computer to install.

Jobs that used to take an hour, a wrench, and a little elbow grease.

Join the discussion, share product recommendations and suggest topics for me to investigate next in the Worse on Purpose community.