In the midst of the chaos that has taken place in the world of Tesla over the last week – with Elon Musk now publicly and repeatedly mocking the Securities and Exchange Commission (and, of course, short sellers) in his latest series of unsupervised tweets - the investing world may have missed that the company released a first-of-its-kind third-quarter vehicle safety report on October 4. Similar reports are going to be released on a quarterly basis going forward.
Since Tesla has already released its production and delivery numbers for the quarter, this seems to be the only piece of information that the company had left to make public aside, of course, from its Q3 financials.
The company's first vehicle safety report appeared to show high marks for the company, and for its semi-autonomous autopilot feature. It revealed that, over the past quarter, Tesla recorded one crash or near miss for every 1.92 million miles driven in its vehicles. That number gets even lower - down to one per every 3.34 million miles – when the vehicle's Autopilot feature was on. Both of these rates, on their face, appear to be far better than the one per every 492,000 miles driven number last reported by the National Highway Traffic Safety Administration.
A resounding success, right? Maybe not.
Immediately, some statisticians and experts pointed out obvious anomalies that led them to believe that Tesla's analysis of its own data is flawed. Put simply, the company has way too small and way too homogeneous of a data set to fairly compare itself to NHTSA averages.
For example, Tesla has only sold under half a million vehicles worldwide. NHTSA data is inclusive of 271 million passenger vehicles in total. This means that Tesla's data set is 0.18% the size of the NHTSAs. Also of concern is the fact that Autopilot is supposed to only be used on limited access highways and the NHTSA total accident figures cover vehicles on all types of roads and conditions.
As if those two differences weren't enough to clearly disqualify Tesla's comparisons, the NHTSA data includes vehicles that have been on the road for years, sometimes decades. Tesla owners are generally more affluent drivers that have much newer cars than average.
In other words, the data sets appear to favor Tesla.
Experts on the subject agree. The Los Angeles Times quoted Nidhi Kalra, a senior technology policy advisor with Sen. Kamala Harris, as saying that “when it comes to injuries and fatalities, we won’t be able to know [the answers] until we’ve had hundreds of millions or billions of miles” of driving history.
Kalra has been on the frontlines of trying to help establish driverless vehicle laws in California and emphasized that "statistics about safety should be used honestly and accurately to reach the best policy decisions."
So, if the results of Tesla's study are difficult – if not outright impossible – to fairly compare to the NHTSA averages, why would Tesla do it? The simple answer - yet another PR tactic - comes to mind.
Additionally, one wonders why Tesla has chosen to start releasing these skewed numbers now, at the end of what is arguably the most crucial quarter in the company's history? After revealing its delivery and production numbers and then watching the stock fall on Friday after CEO Musk's Twitter tantrum, Tesla appears to be running out of quick fixes to the increasingly bizarre behavior of its CEO.