The US Invented Life-Saving Car Safety Ratings. Now They’re Useless.

In the summer of 1980, Jack Gillis, then a 30-year-old employee at the National Highway Traffic Safety Administration (NHTSA), went to visit a vice president at Ford. Sitting in the well-appointed office in the headquarters of a declining titan of the auto industry, in the city of Detroit that was no longer the gravitational center around which all the cars in the world revolved, the vice president told Gillis the same thing all the other car company vice presidents told him: Safety doesn't sell. 

The Ford executive claimed the company didn't come to this position eagerly or lightly. It had studied the issue exhaustively, ever since Robert McNamara—the same Robert McNamara who by the time Gillis graduated from high school was infamous for statistical wizardry that embroiled the United States in Vietnam—tried to market Fords as the safest cars on the road. The company found that consumers say they care about safety when asked, but once they get into the showroom it doesn't factor into their decision. So, in the two decades after that failed attempt to market safety, automakers had arrived at a marketing tautology: They didn't bother to sell safety, because safety didn't sell.

There were fundamental problems with Ford's analysis Gillis didn't know about then—it assumed, for example, that safety differences were the main variable between the cars it studied—but even so, Gillis decided to do his own focus groups. 

Gillis found the Ford executive wasn't entirely off-base, but missed (or omitted) some key nuances. It was true that consumers didn't factor safety into their car-buying decisions, but it wasn't because they didn't care. It was because they made two assumptions, both of which were wrong. 

The first assumption people made was that the government prevented unsafe cars from being sold. This was false. But Gillis, a lowly government bureaucrat, could hardly do anything about that.

The second assumption people made was that all cars were equally safe. They were wrong about that, too. But this was something Gillis could do something about.

Given the U.S.'s recent failures to implement meaningful efforts to address broad systemic health issues, it may be difficult to imagine a bipartisan effort to create a new government agency with the specific mandate of improving public health and safety, all put into effect despite industry protestations that widespread death and injury while going about daily tasks was an unfortunate inevitability. But that is exactly what happened in the late 1960s and early 1970s with motor vehicles. 

If there was a single turning point, it was in 1966 when, after Ralph Nader testified to Congress regarding his landmark book Unsafe at Any Speed, journalists reported that General Motors had been secretly spying on Nader to try and dig up damaging material to attack his credibility. These revelations had the all-too-predictable result of leading lawmakers and ordinary citizens to pay a lot more attention to what Nader had to say. The first federal vehicle safety standards were enacted in 1967 and NHTSA was formally created within the Department of Transportation in 1970 with the mandate to enforce those safety standards. Later, it would also become responsible for fuel economy standards and consumer information programs, too.

Still, NHTSA faced near-constant opposition, especially from the so-called Big Three automakers: General Motors, Ford, and Chrysler. The Big Three dominated the U.S. auto market through the 1970s, accounting for roughly four out of every five cars sold. And they were not used to being told what to do, especially by the government. 

In fact, thanks to their dominance due to an utter lack of international competition following World War II, the behemoth and bloated Big Three were accustomed to telling both the government and consumers what they wanted. Auto companies employed small armies of lobbyists who wandered government agency halls daily, something they could do back in the day. As for consumers, auto companies told them what they wanted, rather than the other way around, a mindset that by the early 1980s was beginning to get them into the sort of trouble that would result in near-constant financial turmoil up through the Great Recession. 

Safety was a perfect example of this dynamic. Aside from the focus groups, NHTSA had other evidence the car companies were wrong about consumer attitudes towards safety. At the time, NHTSA was receiving about 200,000 letters a year from people asking for safety information about cars, according to Joan Claybrook, the head of NHTSA under the Carter administration. NHTSA had the information people wanted because it was legally required to conduct crash tests to ensure cars complied with the few safety regulations that existed at the time. 

Mostly, these regulations had to do with cars "passing" head-on crash tests, in which the cars crashed into a barrier at 30 miles per hour with a dummy, connected with all kinds of sensors, in the driver's seat. Engineers then analyzed the data collected by the dummy to determine how much force it received during impact at the head, neck, chest, and thighs. Engineers had determined how much force each body part could receive before it "failed" the test.

Claybrook could have merely continued the expensive and laborious process of responding to as many of those letters as they could with the detailed, jargon-laden reports the engineers wrote. And in many ways her hands were tied. She couldn't force car companies to make their cars safe enough to pass the tests at higher speeds, or conduct different kinds of crash tests on her own. That would require a time-consuming rule change process or a new law. 

But, Claybrook figured, there was nothing stopping her from doing the tests at a faster speed, a speed at which all cars were not required by law to pass, and simply telling the public how the cars—and the dummies inside—fared. So, Claybrook ordered the crash tests to be done at 35 miles per hour instead of 30. If a car failed at 35, they'd re-test at 30 to make sure it was at least passing the legal minimum. And either way, they would show the public grizzly photos of mangled dummies and car corpses, so they could see what it looks like to be in a car crash.  

Claybrook looked around her staff for someone with a marketing background. She found Gillis, who previously worked in public relations for Western Union, slogging away in the fuel economy department. He immediately got to work on publishing the Car Book, which graded every car NHTSA tested on a pass/fail for safety. He had less than a year to do it before January 1981, when either President Carter would begin his second term or Ronald Reagan would begin his first. To be on the safe side, Claybrook figured it best to release the Car Book before a change in administration.

As I spoke on the phone with Gillis recently, I asked him how the cars did in that first Car Book. The majority, Gillis said as he ran his finger down an old copy, "simply failed." 

Claybrook's program, officially called the New Car Assessment Program (NCAP), was a massive success. Hundreds of thousands of Americans requested copies of the Car Book. 

Rushing the book turned out to be a good idea. Reagan, of course, won the 1980 election on an anti-government platform and his pick for NHTSA administrator, Raymond Peck, unceremoniously killed the Car Book even after Gillis proposed charging two dollars per copy so it wouldn't cost the government a penny. So, Gillis quietly collected the crash test results—which were officially public documents—worked nights and weekends on putting it together into a book, and then quit his job with NHTSA and published it himself. NCAP continued to test cars as it had the year before, it just didn't make the information easily available. Gillis has published the Car Book with the Center for Auto Safety ever since, even after NHTSA once again made NCAP results readily available as they are today. 

In later iterations, NCAP evolved into the star-based crash safety ratings—five being the highest—so ubiquitous in car ads in recent decades. And it had exactly the impact Claybrook and Gillis intended. Automakers started to design their cars so they would pass the tests at 35 miles per hour instead of just 30, therefore making them safer. The companies also competed on whose cars were safer, because getting a bad score hurt a car company's reputation and impacted sales. As a result, car companies worked to get better scores that they could advertise. Buyers shifted preferences within the same class of cars to ones that did better on safety tests, exactly the dynamic auto executives insisted would never happen (and one that, incidentally, initially benefited the Big Three which made bigger, less fuel-efficient cars that tended to do better in those early crash tests than their lighter, more fuel-efficient foreign competitors). American cars were safer than they had ever been before, likely saving thousands of lives every year and preventing tens of thousands of debilitating injuries, all for the cost of publishing information it already had.

The rest of the world noticed. Starting in the 1990s, other regions created their own NCAPs. Today, there are eight NCAPs, one each in Australia, South Asia, China, Europe, Japan, Korea, and Latin America, each with similarly astounding achievements. In Latin America, where car safety regulations are generally weak, the best-selling Chevrolet Onix went from zero to five stars in less than a decade and now includes many safety measures still uncommon in the U.S. market. In its first 20 years, Euro NCAP estimated its program had helped save more than 78,000 lives.

"NCAPs are really incredibly dynamic instruments in vehicle safety," said David Ward who created Euro NCAP and is now secretary general for the Global New Car Assessment Programme that helps create NCAPs around the world. Not only do NCAPs generate a market for safety by informing the public, Ward said, but they also create a competitive environment for car companies that, if they so choose, can show that they are doing more than regulations require. 

In other words, the Ford executive Gillis spoke to 40 years ago had it wrong. Safety does sell, but it requires more than just a few ads in glossy magazines or an executive dropping quotes to a reporter about how the company "values" safety. It requires a neutral third party to track the market and constantly test the products. The game needed a scoreboard.  

"The U.S. deserves a lot of credit for inventing it," Ward said, particularly singling out Claybrook for her ingenuity. 

"It's just a shame," he added, "that the U.S. has now fallen behind where it should be."

The U.S. NCAP has come full circle. The dynamic Gillis identified more than 40 years ago, in which consumers assumed all cars were equally safe, is once again the case. The standards by which NCAP tests cars haven't been upgraded to incorporate new tests or safety technology in more than a decade, resulting in a complete flattening of the grade curve. 

In the 2020 model year, NHTSA crash-tested 353 cars in the U.S. market, according to its website (many of those cars were different models of the same make; for example, NHTSA tested the 2020 Corolla four-door, four-door Hybrid, and hatchback). Of those 353 cars, 73 percent received a perfect five-star rating. The rest received four stars, traditionally regarded as a good enough score to advertise. The only vehicle to receive fewer than four stars was the Ford Transit van. By way of comparison, Euro NCAP looks at the data slightly differently, with the distribution of ratings based on market share instead of cars tested, but seven percent of cars sold received three stars or fewer.

To most car-buyers, this may seem like evidence cars are equally safe and they should base their purchases on other criteria, as Gillis found people did in the late 1970s. But, just as that wasn't true 40 years ago, it's not true now. 

When it comes to good government, it is not enough to pass good laws, enact meaningful regulations, and implement them competently. Like the roads, bridges and tunnels vehicles travel on, good government and consumer protection programs must be maintained over years and decades. Or else it stops working. And that's what has happened with NCAP.

"There are differences between the cars," Claybrook told Motherboard. "But you don't see them in the data. And you don't see them in the data because the standard is so low that all cars comply."

Not only has the U.S. NCAP fallen behind its international peers, the U.S's auto safety landscape is dire, regressing so badly the entire process is in desperate need of reform. And NCAP is a perfect example of why. 

Any way you slice the data, U.S. roads are getting deadlier while those of other developed nations, especially ones with robust NCAPs, are getting safer. In 2018, the most recent year for which NHTSA has published data, the fatality rate on U.S. roads was 11.17 per 100,000 people. In the EU one year later, the rate was 5.1 fatalities per 100,000 people, less than half the U.S's. Between 2010 and 2019, the EU reduced road fatalities by 23 percent. From 2010 to 2018, the U.S.'s road fatalities increased 11 percent.

The charts below break down road fatalities per million inhabitants by various countries since 1995 (for the U.S., the data may vary slightly from the above paragraph which uses official NHTSA statistics while the below charts are Organisation for Economic Co-operation and Development [OECD] estimates). While most countries including the U.S. successfully reduced road fatalities in the mid-2000s, most were able to continue that progress. Not the U.S.

There are many reasons for this divergence. European cities aggressively install traffic calming measures and promote alternative and safer modes of transportation such as cycling and public transport (which were already more common in the EU than the U.S.) Perhaps more importantly, the U.S. has a love affair with large vehicles. Ironically, car companies leveraged the safety-conscious environment ushered in by NCAP to market bigger, more expensive vehicles under the guise that they're safer for the people inside, a claim that is utterly untrue. And without an NCAP that takes all aspects of safety into account, car companies can easily get away with such misleading claims. 

For that reason, one under-appreciated cause of the US's comparative auto safety regression  is that other NCAPs, including those in the EU, Japan, and Australia, continue to be positive forces pushing the auto industry to release progressively safer cars. In the meantime, the U.S. NCAP program has completely stalled.

The U.S. NCAP consists of just five tests: slamming the front of a car into a barrier, slamming a barrier into the side of a car, slamming a pole into the side of the car, and two tests relating to stability and rollover resistance (the latter implemented only after decades of deadly rollover crashes in SUVs). 

Despite all the advances in automotive safety in the last decade, these tests have not changed. Nor has NHTSA introduced any new tests in that time. Lee Vinsel, a professor at Virginia Tech who wrote a book on the history of auto regulation, said manufacturers know exactly what they have to do to ace the test every year, so they "design to the test."

Meanwhile, Euro NCAP continues to demand better from auto manufacturers. The Euro NCAP has 21 different tests, including ratings for rear seat occupants, child seat installations and protection, and for people outside the vehicle like pedestrians, cyclists, and motorcyclists, entire categories the U.S. NCAP completely ignores. While Euro NCAP keeps getting tougher, the U.S. NCAP stays the same.

Not a single one of the more than a dozen experts, current and former NHTSA employees, and safety advocates I spoke to for this article offered any real defense of U.S. NCAP, at least as it is currently constituted. It is, to all observers, the equivalent of a junker, a failing program in need of a major overhaul.

In the final years of the Obama administration, NHTSA tried to do such an overhaul under administrator Mark Rosekind that would have made our NCAP a lot like Europe's. According to Secretary General at Euro NCAP Michiel van Ratingen, his office worked with U.S. NCAP to explain how some of their tests worked, especially pedestrian and offset crash (where cars collide off-center, as tends to happen in the real world more often than direct head-on collisions).

In December 2015, NHTSA filed a request for public comment on these changes, saying it intended to implement them in 2018. But that never happened. The clock ran out on the Obama administration and President Trump never bothered to appoint a successor to Rosekind, making the four Trump years lost ones in terms of auto safety.

It is hardly surprising the Trump administration dropped the ball on a key reform process that began under Obama. But the better—and more relevant question going forward—is not why the clock ran out on the Obama administration, but why the reform process took so long to begin with. 

There is a tendency in a representative democracy to assume that once the government passes a law or issues a new regulation the work is done. But the exact opposite is true, particularly when one political party's platform is to reduce the size and reach of most government programs. In the United States, the government has to proactively and repeatedly fund ongoing programs. Very little is automatic. It has to update regulations through the rulemaking process, which doesn't require Congressional action but takes time and an executive branch committed to those changes. The result is a number of pinch points that can be used to block or delay important upgrades. Over time, if this happens enough, existing programs erode or die in a kind of regulatory shadow banning.

NCAP is a case in point. In truth, the start date of these NCAP reforms was not 2015, but two years earlier when NHTSA first requested comments on overhauling NCAP. This notice, posted in 2013, does as well as anything to encapsulate the absurdity of the federal rulemaking process and how it cannot possibly keep up with technological innovation. Here is the process NHTSA laid out:

In other words, NHTSA, as prescribed by law, undertook a process that is unlikely to occur under a single presidential term, making it vulnerable to party politics that is increasingly about simply undoing whatever the last president did. 

NHTSA says the delay came down to valuable feedback. “NHTSA worked to address the more than 300 public comments received from NHTSA’s 2015 NCAP notice,” a spokesperson told Motherboard. “The commenters raised substantive issues involving both data and test procedures. Vehicle manufacturers, along with safety advocates, cited a need for additional data demonstrating that the program changes would provide sufficient benefits of safety to warrant inclusion in NCAP.  We value the information we receive during our public comment periods and take into account public perspectives as we develop and update our programs and procedures.”

Another pinch point the government has used against crash test programs is the budget. You'd be hard-pressed to name a government program where Americans get more bang for the buck than NCAP. The program has eight full-time staff members, according to During Obama's second term, according to an agency spokesperson. NCAP received approximately $10 million in funding annually, or approximately $250 per death on U.S. roads. The Trump administration actually increased NCAP's budget slightly before slashing it to $8 million in 2020. In other words, the U.S. government spent more money on providing transportation and security for four Trump trips to Mar-a-Lago than it did on NCAP annually. 

All of this results in programs like NCAP stagnating. Similar to a car driven by someone first learning how to use stick, efforts to make it better stall, over and over again, and it lurches forward only to shut off and start all over again.

How do we fix NCAP? It is a question every U.S. resident should be invested in, because it will make our roads safer. And not just for drivers. 

As far as government programs go, NCAP is as unobtrusive as it gets. Consumer advocates love the program because it informs people about something many of them care about. Free market fundamentalists can get on board because it is not a regulation; auto companies could continue to sell cars that get zero stars if they so desired. In fact, NCAP addresses what economists refer to as "asymmetric information," a market inefficiency where one party in a transaction knows more about the product than the other (the literal textbook example of this, by the way, is used cars). Some on the Left criticize such consumer information programs as not forceful or influential enough in actually yielding better outcomes, but NCAPs have often acted as a soft transition towards regulating performance standards. In other words, NCAPs remain an increasingly rare thing: a non-ideological, bipartisan program that actually helps people. At a time when it feels impossible both political parties can agree on something, reforming NCAP is about as close as it gets.

That being said, the Biden administration doesn't need Republicans to sign off on anything. The simplest path forward is for the Biden administration to largely pick up where the Obama administration left the 2015 reforms, dust off the plan, and get it enacted. When asked about the agency’s plans for NCAP, a NHTSA spokesperson told Motherboard, “As new technologies and innovations are incorporated into the U.S. fleet, NHTSA is considering numerous safety additions in the development of a proposal to upgrade NCAP.”

This would be the easiest option, but probably not the best one. As Euro NCAP demonstrates, a lot has changed in automotive safety technology in four years. More cars automatically stop if they detect an imminent crash. Some have similar technology if they're about to hit a pedestrian. Others automatically prevent the car from swerving out of the lane or changing lanes when there's a car in the blind spot. And manufacturers like General Motors and Tesla have varying degrees of driver assist programs that take control of the car, with human supervision, on certain highways. The list goes on. Maybe the 2015 plans need to be updated, once again.

Which brings us to a bolder proposal, to not just overhaul NCAP's tests, but to remove the program from the government entirely so it no longer has to adhere to this meandering rulemaking process. 

Euro NCAP has been so successful, van Ratingen argues, in large part because it is not a government program. It is a member-based not-for-profit in which several European nations' departments of transport are members (automakers and suppliers are not allowed to be members). Van Ratingen feels there's a big difference between NHTSA's performative effort to be collaborative through the request-for-comments on the Federal Register and a true collaboration with advocates, automakers, suppliers, and governments to continuously improve the program. 

Understandably, some advocates are skeptical that working with automakers on auto safety is a good idea, given the industry's historical resistance to anything that even resembles government interference and constantly co-opting the language of safety to sell more expensive cars. But van Ratingen says, because modern cars incorporate so much complex safety technology, any regulation or safety protocol designed without talking to automakers—and especially suppliers who actually invent the new technology in most cases—is doomed to fail. As Van Ratingen put it, "With new technology it’s almost impossible to build protocols from scratch if you don’t know the technology inside and out."

To illustrate the problem, he compared the years NHTSA spent seeking and reviewing comments through the rulemaking process with what Euro NCAP accomplished in that time. In roughly the same time span, Euro NCAP adopted automatic emergency braking testing for rear-end scenarios, expanded it to pedestrian scenarios with an adult walking, running, and a child coming behind the vehicle, then added cyclist and off-center scenarios on top of that, and once again upgraded it to night-time testing and when the car turns across paths or crosswalks. European cars get graded on all of those things now—not just if they have the systems but how well they work—and only if they are standard on all models. And Euro NCAP will keep updating the tests every two years while adhering to a long-term roadmap, an especially important feature considering the gradual transition to cars with more automated driving capabilities.

Meanwhile, all NHTSA will tell you is if a car has optional automatic emergency braking. It won't tell you which models, how much it costs, or how well it works. 

This difference stems from a definitional problem as much as anything. As brilliant as NCAP was for its time, the definition of what a "safe" car is has changed as much as the cars themselves. Instead of simply dusting off the old plans, reforming NCAP entails a broader rethinking of what a "safe" car is in the 21st Century and how to keep that definition from becoming stale. 

Today, NCAP's definition of a "safe" car generally does not include children; adult dummies are used in the majority of tests, and only in the rear seat for the side crash test. It also doesn't include people outside the car, a major reason why SUVs and pick-ups rate so well on NCAP despite them being a major cause of the surging pedestrian and cyclist deaths. Aside from using crash-avoidance technologies like Euro NCAP does, vehicles could be tested on dummies outside the vehicle just as they are on ones inside to rate the force of impact, but they are currently not. Cars can be designed to be safer if they strike pedestrians by using softer materials on lower bumpers and hoods and incorporating more space between the hoods and engine components, and automakers that opt not to do this will suffer lower safety ratings. 

"It’s actually quite strange an NCAP focuses only on car occupants," Van Ratingen said, "and only adults in many cases.”

Likewise, NCAP's definition of "safety" is still restricted to crashes. This was understandable for the 1970s when global warming was barely understood and not linked to automobiles. But that has changed. Vehicle emissions are a major public health issue, both in terms of neighborhood health and the health of the planet as a whole. NHTSA, in conjunction with the Environmental Protection Agency, publishes fuel economy ratings, a practice that first began as a political effort to encourage lower gas consumption during the OPEC oil embargo. 

Perhaps it's time to fold those fuel economy scores into the overall safety rating to create a more holistic picture of how new vehicles impact not just our own safety and those in our immediate vicinity, but also that of the world we live in. 

Even if NHTSA incorporates none of those changes, perhaps the biggest fix could be another definitional tweak. Instead of using the clumsy if approachable five-star score, NHTSA could change the rating system to be more competitive. They could use all that data they collect from crash test dummies to grade on a curve, so, by definition, some cars will always finish in the top 10 percent, some in the bottom 10 percent, and the rest in the middle 80. Nothing about the tests would have to change except for the way the data is processed and presented, and it would ensure we never have the participation trophy scenario we have today. As Gillis and Claybrook showed, it is sometimes the simplest things that make the biggest difference.