Tesla Does Not Have Enough Car Accidents in Sweden for Insurance Company Safety Analysis

Swedish insurance company Folksam releases the results of its annual study on the country’s safest cars. Folksam bases their safety analysis on actual accidents. However, Tesla cars only had seven accidents in Sweden. This is not enough for Folksam to perform statistical analysis.

14 thoughts on “Tesla Does Not Have Enough Car Accidents in Sweden for Insurance Company Safety Analysis”

  1. The title is misleading (not Brians fault), it’s a EU definition thing. IE of the cars that are in crashes, what is their ability to protect the occupants? Not % of accidents/no. cars. The sample size is “cars that have crashed”.

    Ie if there are only 7 crashes (regardless of how many of these cars are on the road), what is the RESULT of the crash, in terms of safety “ability” (e.g., risk of death, injury to occupant etc). The sample size, in Tesla’s case, is 7, which is not enough to determine whether it is a “safe” car or not (when an accident occurs).

    The logic is that cars don’t cause accidents, humans do. There could be ZERO crashes involving a Rolls Royce, but it doesn’t mean Rollers are safer than, say, Hummers or a VW Golf. But if you get t-boned with a Golf, what is the ability of the Golf to protect you from dying (e.g., crumple zones) compared to a similar vehicle? That is the purpose of these safety reports.

    Reply
  2. You’re absolutely right about the basic statistic of “how likely is this car to crash?” If the denominator is big enough it doesn’t matter what the numerator is.

    But I suspect the issue is the full range of other stats that such crash statistics generally calculate:

    • What proportion of Tesla crashes involve the car being written off?
    • What is the average repair cost of Tesla crash damage?
    • What is the repair time of the average Tesla crash damage?

    That sort of thing.
    That needs a minimum number of actual crashes before you can get any decent statistics.

    Then of course, the news goes through 3 different interpretations, including twitter, so by the time we get it the originally coherent story becomes silly.

    Reply
  3. I checked out the pdf. Losses average $3 per car, $12 per Tesla. Time period doesn’t include 2019 so just about all S and X.

    S and X cost 4x as much as the average car, so not much to see there.

    EDIT: Mercedes C Class was worse

    Reply
  4. Statistically, something doesn’t seem right about this article. If the denominator (total number of Tesla’s being driven in Sweden) were high enough then any numerator (# of Tesla crashes per year) including zero crashes should give a result with high enough statistical confidence to be useful for insuring. For example, if there were 10 million Teslas being driven each year with zero crashes wouldn’t any insurance company want to insure them? The larger the denominator the less likely a random variation in the numerator would make a substantial difference in the ratio of crashes per total # of Tesla’s per year.

    Reply

Leave a Comment