greetings all. i realize i haven't posted in a while. i've been doing a lot of traveling
to my numerous acquaintances on the board, i apologize that my vacations have kept me from doing this analysis sooner, as i'm sure it would have saved many people much money since the third fire was reported. but hey, you get what you pay for.
i felt like now is a good time to share some of my insights, specifically regarding the tesla fires. i know many people won't like to hear what i have to say. and i don't have a longer term holding in tesla any more, although sometimes i day-trade tesla from the long or short side. tesla has been kind to me, both as an investor and model s owner. so although i have been very tempted to short tesla the past months, i just can't bring myself to establish a meaningful short against the company that has been so good to me as a customer and investor. who knows, maybe i will lose my sentimentality and return to being the cold-blooded capitalist i usually am.
as i mentioned, my perspective is that of a model s owner, a former investor, and potential short or potential re-investor. i've been lurking on the forums and i've seen some good statistical information posted and discussed. my purpose here is to present the various points of view, point out the subtle differences, and present statistical data to support. i will make some estimates and some speculations based on the best information i have. i'm sure the collective intelligence here will be able to refine and correct whatever i post to come up with a clearer picture of the truth. or they'll just flame me!
this got too long, so for brevity, i dealt only with vehicle fires per time on the road. you could dig out data and do similar comparisons for mileage which should reach similar conclusions.
so let's start with tesla's point of view, as presented by elon's blogs.
Model S Fire | Blog | Tesla Motors
elon wrote: "The nationwide driving statistics make this very clear: there are 150,000 car fires per year according to the National Fire Protection Association, and Americans drive about 3 trillion miles per year according to the Department of Transportation. That equates to 1 vehicle fire for every 20 million miles driven, compared to 1 fire in over 100 million miles for Tesla. This means you are 5 times more likely to experience a fire in a conventional gasoline car than a Tesla!"
The Mission of Tesla | Blog | Tesla Motors
and he later wrote: "Since the Model S went into production last year, there have been more than a quarter million gasoline car fires in the United States alone, resulting in over 400 deaths and approximately 1,200 serious injuries (extrapolating 2012 NFPA data)... There are now substantially more than the 19,000 Model S vehicles on the road that were reported in our Q3 shareholder letter for an average of one fire per at least 6,333 cars, compared to the rate for gasoline vehicles of one fire per 1,350 cars. By this metric, you are more than four and a half times more likely to experience a fire in a gasoline car than a Model S! Considering the odds in the absolute, you are more likely to be struck by lightning in your lifetime than experience even a non-injurious fire in a Tesla.
Those metrics tell only part of the story. The far more deadly nature of a gasoline car fire deserves to be re-emphasized. Since the Model S went into production mid last year, there have been over 400 deaths and 1,200 serious injuries in the United States alone due to gasoline car fires, compared to zero deaths and zero injuries due to Tesla fires anywhere in the world."
my executive summary of analyzing tesla's statements is as follows: there are numerous inaccuracies, exaggerations, and half-truths in elon's statements, which i will highlight below. despite so many glaring problems with his presentation, the fact is evidence supports his overall assertion (and the key word here is "overall") that the model s is safer than gasoline cars from the standpoint of catching fire. and as a consequence, the overall assertions of lower risk of death from fires, etc. are also valid.
my final executive summary is that even though tesla's overall statements are accurate, they have a major problem developing here because the statistics are clearly indicating something is wrong with the undercarriage. at least as of now, the statistical evidence is that there are too many collision related fires to indicate otherwise.
so let's start with analyzing tesla statements from elon's blogs.
1. "The nationwide driving statistics make this very clear: there are 150,000 car fires per year according to the National Fire Protection Association"
this statement is false. here is the source document from the nfpa.
http://www.nfpa.org/~/media/Files/Research/NFPA reports/Vehicles/osautomobilefires.pdf
i'm guessing pressed for time elon just read the summary introduction on page 1, which says: "During 2006-2010, U.S. fire departments responded to an average of 152,300 automobile fires per
year". i believe elon is quoting this statistic, which then leads to an errant overgeneralized statement. he could have correctly said, between 2006-2010 there were an average of 150,000 car fires per year.
is this relevant? in my opinion it is a relevant omission. see figure 6 in the nfpa document on page 9. you can see a trend lower in automobile fires. the 2006-2010 average is a little above the 2008 data. the most recent data from the document is the 2010 figure that there were 135,000 car fires.
considering how safety technology changes and reporting changes and also considering the strong trend in the data, the most relevant compare is the most recent data one can find. using 5 year old data to make a point comparing safety of new cars today is kind of silly in my view.
however even the 135,000 figure from 2010 is not the correct one to use in my opinion. flip up to pages 6-7 of the nfpa document and you'll learn that 10% of the fires are set intentionally (arson anyone? haha). including these in roadworthiness comparisons is not valid. removing an estimated 10% of fires that are intentional, you end up with about 121,500 automobile fires in 2010 from non-intentional causes.
and even that number is not correct, because the trend of increasing safety has been so strong. referencing another source document (this is the original source for the summary sheet elon uses as a source in his second blog):
summary cited by elon:
http://www.nfpa.org/~/media/Files/Research/Fact sheets/FireLossFacts.pdf
original source:
http://www.nfpa.org/~/media/Files/Research/NFPA reports/Overall Fire Statistics/osfireloss.pdf
scroll down to page 50 in that original source osfireloss.pdf. you'll see a table of highway vehicle fires by year. this data is slightly different from the automobile fire data because it includes a lot of different kind of vehicles, not just cars: "Highway vehicles include any vehicle designed to operate normally on highways, such as automobiles, motorcycles, buses, trucks, and trailers, but not manufactured homes on foundations." note that "highway vehicle" means a vehicle designed to be on a highway, not where the fire happened. the table shows the number of highway vehicle fires went from 184,500 down to 172,500 between 2010 and 2012, a drop of 7%.
so trying to bring the prior 121,500 estimate of car fires from 2010 to 2012, how? i am adjusting it lower 7%, and so i get about 113,000 non-intentional automobile fires in 2012. this is a far cry from the 150,000 figure elon is quoting.
since this figure is so central to making valid comparisons, it's worthwhile to double check it via an alternate estimation method. one way to do that is to combine the annual automobile fire data from figure 6 here:
http://www.nfpa.org/~/media/Files/Research/NFPA reports/Vehicles/osautomobilefires.pdf
and the tabular data from page 50 here:
http://www.nfpa.org/~/media/Files/Research/NFPA reports/Overall Fire Statistics/osfireloss.pdf
| all auto fires | hwy vehicle fires | ratio |
2002 | 230,000 | 307,000 | 0.749 |
2003 | 216,000 | 286,000 | 0.755 |
2004 | 203,000 | 266,500 | 0.762 |
2005 | 189,000 | 259,000 | 0.730 |
2006 | 177,000 | 250,000 | 0.708 |
2007 | 162,000 | 227,500 | 0.712 |
2008 | 148,000 | 207,000 | 0.715 |
2009 | 140,000 | 190,500 | 0.735 |
2010 | 135,000 | 184,500 | 0.732 |
what you see here is 70.8-76.2% of highway vehicle fires are automobile fires. taking the 2012 vehicle fire data (172,500) and applying the range (70.8%-76.2%) results in approximately 122,100-131,400 automobile fires from all causes. taking out 10% of that for the intentional fires gives you between 109,900 and 118,300 non-intentional automobile fires in 2012.
the estimate i derived of 113,000 is within this range. the best thing would be to get the original data, which i don't have. but i believe my 2012 estimate of 113,000 non-intentional automobile fires per year is much more accurate than elon's claim of 150,000 automobile fires per year.
my estimate for total automobile fires puts back the 10% that were intentional - 113,000 / 1.1 = 125,500 total automobile fires.
2. "... and Americans drive about 3 trillion miles per year according to the Department of Transportation. That equates to 1 vehicle fire for every 20 million miles driven, compared to 1 fire in over 100 million miles for Tesla. This means you are 5 times more likely to experience a fire in a conventional gasoline car than a Tesla!
i said i was going to skip the miles driven analysis, but i think i better say a little something since elon did. elon is getting the 3 trillion miles driven from this federal highway administration source:
Table VM-2 FHWA
however this is again an inaccurate comparison, as it includes miles driven by semis, tractors, tankers, and all kinds of vehicles that aren't automobiles. you see my point here? he's trying to use the number of automobile fires and then dividing with total miles driven for **all** vehicles. his estimate of automobile fires is too high and his estimate of miles driven is too high. the result is that the calculation is complete garbage. the numerator and denominator are both too high, meaning the end result is ambigiously high or low. for example, if the real answer is 1/2 and one estimates numerator and denominator high, one might estimate 2/6 or 5/7 - one is too low the other too high with no predictability.
one could try to adjust the mileage data using miles driven by automobiles vs. trucks and numbers of the same from tables 58 & 60 here:
EIA - Annual Energy Outlook 2013 Early Release
however this path is also difficult because the nfpa data defines "automobiles" to include suv's and minivans, and i think the federal highway data throws those into the truck category.
another problem is that the cumulative fire statistics include very old cars, and those have different safety features and therefore fire risk than new cars. all teslas are very new (1.5 years or less), so obviously the direct compare on model s safety is with new cars.
even with all these caveats, i could make various adjustments as follows:
a. get the total miles driven by light duty vehicles in table 60 here:
EIA - Annual Energy Outlook 2013 Early Release
b. take that figure (2658 billion) and try to adjust for the number of automobiles vs trucks by the nfpa definition. my estimate is that automobiles would be driving somewhere between 1500 billion and 2500 billion miles each year
c. now compare the 113,000 non intentional automobile fires for 1500-2500 billion miles driven to what is now 3 auto fires in teslas for about 140,000,000 miles driven. adjust the auto figure lower by about 20-25% to account for fewer fires in newer cars vs. older cars.
the net result i find is that even with 3 fires, apples-to-apples a fire is 1.5x-2.5x more likely to occur in an ice automobile vs. a tesla model s. this broadly supports elon's claim that in terms of fires overall, model s is safer. but it is a far cry from the 5x greater likelihood of fire that was claimed in the first blog.
of course with 3 times as many fires since at the writing of the first blog, you could say elon's 5x figure might be divided by 3 resulting in an ice auto being 1.67x more likely to catch fire than a model s, based on miles driven. so again falling in that range of 1.5x-2.5x that i derived via a different, hopefully more accurate approach.
i've spent more time on the miles driven than i wanted, but you can see why i want to avoid this method of analysis. the salient points i take away are that (a) elon's 5x less likely to catch fire claim is of exaggerated magnitude, and (b) by all reasonable estimates based on miles driven, model s is 1.5x-2.5x safer than ice cars, so the directional claim made by elon is still supported by the data.
3. "Since the Model S went into production last year, there have been more than a quarter million gasoline car fires in the United States alone, resulting in over 400 deaths and approximately 1,200 serious injuries (extrapolating 2012 NFPA data)... By this metric, you are more than four and a half times more likely to experience a fire in a gasoline car than a Model S! Considering the odds in the absolute, you are more likely to be struck by lightning in your lifetime than experience even a non-injurious fire in a Tesla."
the first part of the statement is false, but it ends up not to matter. the source document that is linked is:
http://www.nfpa.org/~/media/Files/Research/Fact sheets/FireLossFacts.pdf
and the statement within is
"172,500 highway vehicle fires occurred in the U.S. during 2012."
but this is "vehicle fires" not "car fires" which is the term elon has used. the distinction is important because vehicle fires incorporates buses, trucks, motorcycles, etc. and doesn't give an apples-to-apples comparison with automobiles (cars).
model s went into production (i think) in may 2012. it's been 18 months = 1.5 years since then, so taking 172,500 annual fires x 1.5 years = 258,750 vehicle fires since model s went into production. that's where elon gets his inaccurate claim that there have been a quarter million car fires in the united states alone since model s went into production.
it's odd that elon is contradicting his prior blog here, in which he said there are "150,000 car fires per year" (still wrong, but more accurate).
trying to do a correct analysis, i start from section 1 above where i provided 113,000 as the 2012 estimate for the number of non-intentional automobile fires. multiplying by 1.5 years gets to about 170,000 car fires since model s went into production, not 250,000. but actually when model s went into production isn't relevant. what's relevant is how many model s are on the roads and for how long.
there were 128.1 million cars on the road in 2012, from table 58 here:
EIA - Annual Energy Outlook 2013 Early Release
113,000 fires/year / 128.1 million cars = .000882 non-intentional fires/car-year
the units are important, because the translation is that there are .000882 fires for each car that's on the road for a year.
now let's see how many teslas have been on the road, and for how long. the little table below takes deliveries by quarter including an estimate for the recent period. over in the far right i estimate the number of car-years on the road for the model s. you can see the total to today is about 13,300 model s car-years.
quarter | deliveries | cum vehicles | car-years |
9/30/2012 | 250 | 250 | 63 |
12/31/2012 | 2400 | 2650 | 663 |
3/31/2013 | 4900 | 7550 | 1,888 |
6/30/2013 | 5150 | 12700 | 3,175 |
9/30/2013 | 5500 | 18200 | 4,550 |
11/20/2013 | 3074 | 21274 | 2,973 |
| 11/20 data estimated |
| |
total model s car-years on the road | 13,310 | | |
for model s we have 3 fires / 13,300 car-years = .000226 non-intentional fires/car-year
doing the analysis this way, i find that the risk for all non-intentional fires is roughly 4x greater in ice vehicles than it is for model s. so again, despite elon's mistakes, the overall statement that model s is much less likely to catch fire than ice cars is clearly supported by the data. again, keep in mind that key word - "overall".
4. "Considering the odds in the absolute, you are more likely to be struck by lightning in your lifetime than experience even a non-injurious fire in a Tesla."
this statement is false and just plain ridiculous. why elon even pulls out this garbage which doesn't even have anything to do with his case i have no idea... i guess it goes to what others have said that p.r. needs a lot of help. and probably he's so busy he doesn't really have time to do all the fact verification himself.
i don't care about odds in the "absolute" and i don't care about being hit by lightning. as a tesla owner, i care about what are the odds i'll experience a fire. the real odds as i derived above are .000226 model s fires per car-year, or about 1 in 4,430... per year!
but just to be complete, wikipedia has the odds of being struck by lightning in your *entire lifetime of 80 years* at 1/6,250. i think this is where elon is getting the faulty comparison of comparing one year or less of model s risk to 80 years of lifetime risk.
Lightning strike - Wikipedia, the free encyclopedia
the source for wikipedia is the national weather service.
NWS JetStream - Lightning Frequently Asked Questions
here you learn the odds of being hit by lightning in a given year are 1 in 500,000. apples-to-apples, one year-to-one year comparison, it's over ** 100x ** more likely that my model s will catch fire than it is that i will be struck by lightning.
ok, and now let's move on to why i think tesla has a major problem developing here. i think dm33 on this board has brought up this point a few times, but he hasn't been clear enough to really drive home why there's a problem. that's what i will try to do here.
let's circle back to the 2012 estimate of the number of car fires from section 1: rehashing...
"taking the 2012 vehicle fire data (172,500) and applying the range (70.8%-76.2%) results in approximately 122,100-131,400 automobile fires from all causes."
and the key here is that this is the statistic for "all causes". delving deeper, we can review figure 2 from this source document:
http://www.nfpa.org/~/media/Files/Research/NFPA reports/Vehicles/osautomobilefires.pdf
what we see here is that almost 70% of car fires are from mechanical or electrical malfunction. only 4% of fires are from collisions.
we'll use my prior estimate of the number of total automobile fires from section 1: 125,500.
so estimating the total number of automobile fires per year from collisions, we would get: 4% x 125,500, or 5,020.
5,020 collision related car fires per year / 128.1 million cars = 0.0000392 collision related fires per car-year. this number is probably too high for an apples-to-apples comparison because the 5,020 collision related fires includes many older cars, adjusting for age
now note that all the tesla fires were collision related. from section 3 above, i derived the similar statistic for tesla model s:
"for model s we have 3 fires / 13,300 car-years = .000226 non-intentional fires/car-year"
notice where the decimal points are - there's almost an order of magnitude difference here. in fact, my best estimate is that the model is 5.75 times *** more likely *** to catch fire in a collision than a standard ice automobile.
sadly, this number is probably too conservative for an apples-to-apples comparison because the 5,020 collision related fires includes many older cars and the teslas are all brand new. adjusting for age, i'd guess the real number is probably that the model s is 7 times more likely to catch fire in a collision than a similarly new automobile.
ok, brace yourself for the statistical punch line. from an investor's perspective, the real question is what are the odds we would see 3 collision related tesla fires in 13,300 car-years on the road if in fact the tesla was as collision-related fire-safe as ice cars?
this can be approximated using a binomial or normal distribution.
use p = 0.0000392 collision related ice fires per car-year for the probability of seeing a collision related car-fire
use 3 for the number of fires (successes in terms of the binomial distribution)
use 13,300 for the number of observations in the binomial distribution
what we want to know is under the assumption that the mean is .0000392, what are the odds of seeing 3 or more fires in a batch of 13,300 car-years (the model s "n").
the answer is 2.1%. meaning, even at the 97.5% significance level we can reject the notion that the model s is as fire-safe as an ice car in a collision. statistically, it's almost highly probable there's a serious problem here.
i'd be shocked if the nhtsa didn't already realize this. i'm not sure if tesla has this figured out yet, but my guess is someone at tesla has been or will soon be thinking really hard about what's going on.
please note that in all statements made, one must be very clear in the choice of words to be accurate. are you talking about vehicle fires or car fires? are you talking about all non-intentional car fires or just collision related car fires?
the reason this issue turns into a mess in the media is most journalists get confused by all the various subtleties in the data and are unable to come to a valid conclusion. as an investor, i make my living this way so i have to do my very best to to get it right.
the key to my whole conclusion is the 5,020 estimated automobile collision fires per year. if anyone can prove that number is far off, please do it and show me in this thread. i've tried it a bunch of different ways, and that's what i keep coming up with, and it's not a tesla friendly answer.
do not duplicate this research in another location without sourcing the original post.