Is that a plane? Superman? No, it's likely Elon Musk milliseconds after his latest tweet.
There is an interesting controversy again brewing in Tesla (NASDAQ:TSLA). It concerns the claim that the Model 3 was the safest car ever tested by the NHTSA. Tesla made that claim, and the NHTSA rebuked it.
The reason why it rebuked the company’s claim is immediately clear. The NHTSA rates cars’ safety on a 5-star scale. The Model 3 got 5 stars, but so did many, many other cars. As the NHTSA said (bold emphasis is mine):
The National Highway Traffic Safety Administration’s New Car Assessment Program (NCAP) - known as the 5-Star Safety Ratings Program - provides consumers with information about the crash protection and rollover safety of new vehicles beyond what is required by Federal law. NCAP conducts a total of three crash tests on new vehicles: one frontal and two side crash tests, as well as a rollover resistance assessment - a driving maneuver test that assesses a vehicle’s susceptibility to tipping up and a measurement of how top-heavy a vehicle is. Results from these three crash tests and the rollover resistance assessments are weighted and combined into an overall safety rating. A 5-star rating is the highest safety rating a vehicle can achieve. NHTSA does not distinguish safety performance beyond that rating, thus there is no "safest" vehicle among those vehicles achieving 5-star ratings.
However, Tesla’s claim is somewhat different. You see, the company isn’t claiming that the Model 3 is actually "the safest car ever tested by the NHTSA". Instead, it claimed that:
As a result of this slightly different claim, it turns out that both the NHTSA and Tesla can be right at the same time:
- On one hand, the NHTSA does not differentiate between 5-star rated cars, and there’s no 6-star (or even 5.x star) rating. So, none of those 5-star cars can claim to be safer than another 5-star rated car.
- On the other, the NHTSA actually does publish detailed data on the crash tests it performs to arrive at the “stars” rating. And these tests are indeed then computed into an overall probability of injury. And therein lies the reason why Tesla is right. According to the test details, and with a single caveat*, the Model 3 is indeed the car with the lowest probability of injury, and thus, arguably, the safest car ever tested by the NHTSA.
Bear with me, I’ll go into more detail now.
A Particular Breed Of Safety
First off, what kind of safety are we talking about? We’re not talking about overall car safety here. There are two broad “safety categories” when it comes to a car:
- Active safety. This refers to the features a car possesses whose intent is to prevent an accident. Think ABS, ESP, tires, lights, roadholding, etc.
- Passive safety. This refers to the features whose intent is, after a crash occurs, to minimize injury to its occupants. Think about crash structures, survival cells, seat belts, airbags, etc.
When it comes to the NHTSA tests, those focus on passive safety. The NHTSA does make recommendations regarding components which help active safety, but it can’t take those into account when establishing a probability of injury. Instead, starting from standardized crash tests, and using sensitive crash test dummies, the NHTSA carefully monitors the effects of each crash test on several points in each crash test dummy.
It is from the severity of the potential injuries to each body part in each standardized crash test that the NHTSA will derive an overall probability of injury.
Hence, this means that Tesla claiming a lower probability of injury during the NHTSA tests translates into the crash test dummies involved in those tests actually having suffered less significant aggregate injuries in the Model 3 than in other cars. In other words, in as much as those tests replicate real-life crash consequences, the Model 3 likely does indeed expose its occupants to a lower injury likelihood versus another car, all other things being equal*.
As it turns out, one can even check out the detailed results of NHTSA’s tests, including by test and body part. For instance, here are the results, which include the Model 3 along with several other cars.
So, why can the NHTSA say that the results don’t allow to say one car is superior to another? Well, a crash is a chaotic thing. If the same car is crashed twice, the results might well differ to a degree. Plus, the tests are standardized, whereas real-life crashes have infinite variations. So, a car that performs better than another one in a given standardized test might well perform worse in a real-life crash that differs only a bit from the standardized test.
As a result of this, the NHTSA allocates cars to broad buckets (the stars). Within those buckets, it expects the cars to broadly perform better than in an inferior bucket, but feels the uncertainty implied within these tests and real-life variability is such that it can’t differentiate one car from the next. Think of a drug that seems a bit better than placebo, but whose statistical confidence isn’t high enough to say that it actually works. A rather similar (but much more uncertain) phenomenon is at work here.
However, in the 5-star bucket (as well as on the 1-star bucket), this can create weird conclusions. For instance, think of a hypothetical car where it is impossible to get injured. That car would still get a 5-star rating. No one could then claim this 5-star car was the safest, even though no other car could avoid injury altogether. This is absurd, and here the Model 3 is exposed to a similar (but not as extreme) dynamic. The car excels in the crash tests to an extent the others don’t, yet it can’t formally claim to be the safest.
Anyway, in my view, the NHTSA is right regarding the stars system, but Tesla is right in pointing out that the Model 3 did indeed outperform all other cars on the NHTSA’s crash tests*.
The Data Confirms It
Tesla had already made similar claims in the past with the Model S and Model X. We can see, though, that those had a lower probability of injury by a smaller margin than the Model 3.
Did this probability of injury translate into real life? First, one should remember that Teslas are getting a reputation of being expensive to insure. If they’re expensive to insure, the reason has to be that paying claims on Tesla crashes is also expensive.
These claims broadly cover material damage (fixing the car, property damage) and injury (personal injury compensation, medical payments, bodily injuries). So, how can Teslas be expensive to insure if they’re safer than other cars? There is an amazing answer:
- On one hand, Teslas crash more often (I, II) than other cars, and it’s extremely expensive to fix them. This results in extremely high claims for collision damage and somewhat higher claims for property damage (since they crash more often).
- But on the other, the Teslas have much lower than average claims for injuries, especially if you exclude injuries to others!
This is borne by the insurance claims data; for instance, here is the page on the 2015-2017 large luxury cars cohort. Notice how the Tesla Model S has extremely high collision costs, yet extremely low personal injury and medical payment costs:
Source: IIHS. The results are adjusted for operator age, calendar year, density, gender, marital status, model year, risk (standard or nonstandard) and state.
The low injury losses are to be even more emphasized when you consider that Teslas crash more often than other similar cars (“crash frequency” is a reason given for insurance rate increases, though exact data cannot be found for it - check links above for examples).
Is the Tesla Model S the one with the absolute lowest injury claims? Not really, several large pickups and SUVs, along with the Porsche 911, tend to score better. But if we are restricted to cars, the model does tend to come up on top (and again, it is to be expected that the data is unfavorably affected by crash frequency).
Still, we can safely say that real-world data is reflecting the NHTSA’s probability of injury, thus giving credence to Tesla’s claim.
There’s an important caveat to be included in several places in this article. As per the NHTSA, its probability of injury in part of the crash tests (the frontal crash test) is not directly comparable among different car models. Instead, it’s just comparable for cars with similar weights (+-250lbs).
Thus, it’s possible (and indeed, seems to be borne by real-world data) that a higher weight will make some cars safer in practice. This is the reason why several pickups and SUVs “score well” in real-life injury losses. They do so at the cost of other, less-massive road occupants in frontal crashes.
I have two conclusions here:
- Tesla’s claim of being the safest car ever tested by the NHTSA is, in my opinion, “true enough”. There are caveats, and the NHTSA itself doesn’t want to go beyond the 5-star system. But truth is that the crash tests provide more detailed data, and that data favors the Tesla Model 3.
- Tesla is to be commended for having produced cars which test so well in crash tests, and whose passive safety also seems to translate well into the real world. This is in spite of several fatalities affecting Teslas, in excess of what one would find in many competing vehicles. One could argue that “driven the same way”, the crashes a Tesla Model 3 would get involved in would, on average, result on lower injuries to its occupants.
Commending Tesla on these grounds is not to be taken lightly. One of the areas most new car business entrants seem to have most trouble with is passive safety. The Japanese and Korean brands had problems in this area for a long time. And the Chinese, nowadays, are still finding this area hard while they catch up on so many others. Tesla, from its first volume car (the Model S), had this area at a very high standard. That’s impressive.
Safety can be an important selling argument favoring the Tesla Model 3 (and Teslas in general, as long as driven sensibly).
Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours.
I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.