The VIN Gap - a Hypothesis
[Updated 2/6/2019 with fresh legal data. See below]
If you're a Tesla bear and haven't read Bad Blood, well stop reading this and go get yourself a copy. It's a great read, but I suspect in a year or two when "Disorder of Magnitude- The Elon Musk Story" (h/t @ElonBachman) comes out it'll make Theranos seem like Sesame Street.
In the wake of the SEC & 35K announcement many bears think the end is nigh. I'm skeptical of that, but it's a good time to clarify my thoughts and put down my guesses as to what's really going on at Tesla. Hopefully in a year or two, we can look back and see how accurate or wildly inaccurate my inferences were. I should add that my track record here on what's going with Tesla is not great, but you only hit the pitches you swing at.
In this, the first of two posts, I'll explain the "VIN Gap", what it means, and why I think Tesla has acted as it has.
A group of bears has been tracking US Model 3 registrations. I won't go into details, but suffice it to say, this data source covers about 97-98% of all US Model 3s. Starting in Sept we noticed a huge gap between the numbers Tesla claimed to have delivered and the #'s showing up as registered. This gap became quite large- at one point more than 30K cars were missing. Even now, more than 2 months into the new year, there are still about 2K fewer M3's registered than were supposedly delivered in 2018. And we know several thousand Jan & Feb cars have already been registered.
Since the end of the year, the gap has shrunk as more data has come in, but there seems to be a persistent 5-6K gap between Tesla's reported deliveries and US registrations. Either Tesla is lying about deliveries or 5,000 M3's are really unregistered.
Hard evidence for this delay comes from two sources. First, NY publishes registrations on a monthly basis. It was often the case in Q4 that less than 1/2 to 1/3 of M3 registrations showed up in the month. That is when NY released Sept data, only 1/3 of the supposedly sold M3's were registered. A month later when they released the Oct data, the number of M3's registered in Sept had doubled! This implies that Tesla was very slow with registering cars. No other OEM exhibits this behavior.
Further evidence of the slowness in registration comes from user-survey data, where we can see the time lag between when a user reported delivery of an M3 and when that VIN showed up as registered. It was usually tracking at a minimum of 3 weeks, often 6-8 weeks of delay.
A third, anecdotal source is the various Twitter reports of expired temp tags, tickets for expired tags. Common among Tesla owners, but virtually unheard of for other new cars.
It's also hard to explain the delivery numbers. It's not Elon's m.o. to lie outright, so there must be someway to square the delivery numbers with the registration numbers.
The "Clean Title" hypothesis nicely explains both these phenomenon.
Tesla offers allow you to return your car for any reason if you haven't take a test drive. We know from Elon that ~70% of Model 3's were bought without a test drive. In addition, we know that the initial quality on many M3's was so low, buyers returned them either outright or for new cars without defects.
We'd expect to see a couple thousand "lightly used" M3's for sale, having been driven of the lot by a new owner, then returned. However, as far as anyone can tell, Tesla hasn't sold a single "used" M3.
So where are all these returned M3s? I believe that they've sold these cars as high-mileage "Demo" cars.
Here is where the registration delay comes in. By delaying registration for 3-4 weeks they ensure that the titles remain "clean" when the car is returned. No state record of the car being sold. Now they can turn around and sell the car as new (but with a few hundred/thousand miles on it).
The smartish bears will also see how the delivery gap is explained. When they resell the returned car, they've now "delivered" the same car twice! This is how 60K "deliveries" might translate to only 57K cars on the road- if 3,000 of those cars were delivered twice.
As a side note- if you've returned an M3 in Q3 or Q4, please send the VIN in the comments or on twitter (@quirkyllama). We could confirm this hypothesis by checking a VIN known to be returned and seeing if the title history properly reflects it's twice-sold status, or if Tesla "erased" the previous sale and sold the car again as new.
Update- 2/6/2019
As I suspected, selling a previously sold car as new is illegal.
Badger24 provides this handy screenshot of FTC regulations.
If you're a Tesla bear and haven't read Bad Blood, well stop reading this and go get yourself a copy. It's a great read, but I suspect in a year or two when "Disorder of Magnitude- The Elon Musk Story" (h/t @ElonBachman) comes out it'll make Theranos seem like Sesame Street.
In the wake of the SEC & 35K announcement many bears think the end is nigh. I'm skeptical of that, but it's a good time to clarify my thoughts and put down my guesses as to what's really going on at Tesla. Hopefully in a year or two, we can look back and see how accurate or wildly inaccurate my inferences were. I should add that my track record here on what's going with Tesla is not great, but you only hit the pitches you swing at.
In this, the first of two posts, I'll explain the "VIN Gap", what it means, and why I think Tesla has acted as it has.
The VIN Gap
Source: @TeslaCharts |
A group of bears has been tracking US Model 3 registrations. I won't go into details, but suffice it to say, this data source covers about 97-98% of all US Model 3s. Starting in Sept we noticed a huge gap between the numbers Tesla claimed to have delivered and the #'s showing up as registered. This gap became quite large- at one point more than 30K cars were missing. Even now, more than 2 months into the new year, there are still about 2K fewer M3's registered than were supposedly delivered in 2018. And we know several thousand Jan & Feb cars have already been registered.
Since the end of the year, the gap has shrunk as more data has come in, but there seems to be a persistent 5-6K gap between Tesla's reported deliveries and US registrations. Either Tesla is lying about deliveries or 5,000 M3's are really unregistered.
Registration Delayed is Registration Denied!
In addition to the VIN Gap, we also noticed that Tesla was extremely slow to register new M3s. When you buy a new car, in most states the dealer will have registered the car within 24 hours, and it'll show up in national databases within a week or two at the most. From Sept to Jan we noticed that Tesla was taking 4-8 weeks on average to register new cars.Hard evidence for this delay comes from two sources. First, NY publishes registrations on a monthly basis. It was often the case in Q4 that less than 1/2 to 1/3 of M3 registrations showed up in the month. That is when NY released Sept data, only 1/3 of the supposedly sold M3's were registered. A month later when they released the Oct data, the number of M3's registered in Sept had doubled! This implies that Tesla was very slow with registering cars. No other OEM exhibits this behavior.
Further evidence of the slowness in registration comes from user-survey data, where we can see the time lag between when a user reported delivery of an M3 and when that VIN showed up as registered. It was usually tracking at a minimum of 3 weeks, often 6-8 weeks of delay.
A third, anecdotal source is the various Twitter reports of expired temp tags, tickets for expired tags. Common among Tesla owners, but virtually unheard of for other new cars.
The Clean Title Hypothesis
It's hard to explain the registration delays. It could be a cash grab- Tesla was holding onto tens of millions of dollars in registration fees and state sales tax. I've no doubt that this was an ancillary benefit but it seems like a pretty extreme move, especially given subsequent indications that in Q4 the cash balance was low, but not that critical.It's also hard to explain the delivery numbers. It's not Elon's m.o. to lie outright, so there must be someway to square the delivery numbers with the registration numbers.
The "Clean Title" hypothesis nicely explains both these phenomenon.
Tesla offers allow you to return your car for any reason if you haven't take a test drive. We know from Elon that ~70% of Model 3's were bought without a test drive. In addition, we know that the initial quality on many M3's was so low, buyers returned them either outright or for new cars without defects.
We'd expect to see a couple thousand "lightly used" M3's for sale, having been driven of the lot by a new owner, then returned. However, as far as anyone can tell, Tesla hasn't sold a single "used" M3.
So where are all these returned M3s? I believe that they've sold these cars as high-mileage "Demo" cars.
Here is where the registration delay comes in. By delaying registration for 3-4 weeks they ensure that the titles remain "clean" when the car is returned. No state record of the car being sold. Now they can turn around and sell the car as new (but with a few hundred/thousand miles on it).
The smartish bears will also see how the delivery gap is explained. When they resell the returned car, they've now "delivered" the same car twice! This is how 60K "deliveries" might translate to only 57K cars on the road- if 3,000 of those cars were delivered twice.
As a side note- if you've returned an M3 in Q3 or Q4, please send the VIN in the comments or on twitter (@quirkyllama). We could confirm this hypothesis by checking a VIN known to be returned and seeing if the title history properly reflects it's twice-sold status, or if Tesla "erased" the previous sale and sold the car again as new.
Update- 2/6/2019
As I suspected, selling a previously sold car as new is illegal.
Badger24 provides this handy screenshot of FTC regulations.
Comments
Uhhhhhhh...
- Feb 2018, car is bought and delivered to Dallas, Texas to buyer.
- Buyer rejects delivery due to damage to the trunk.
- May 2018. Car was still on the Tesla prep lot.
- May 2018, TMC User PcPro is assigned the VIN but shortly thereafter Tesla "unassigns" it.
No further idea of what happened to it after that. Would be interesting to know when and where it was registered. Can you do that?