It’s been a bad year for Tesla so far. The stock is way down and so are new car sales. But there’s another problem that has received less attention.
The auto insurance business is in deep trouble.
The last time I wrote about Tesla selling auto insurance, I suggested it was a terrible idea. After looking at their 2023 results, it is more clear than ever that I was correct.
As a reminder, last time I suggested three reasons Tesla shouldn’t sell insurance.
- It increases operational and financial risk as it is not financially diversifying.
- If it doesn’t go well, it hurts Tesla’s reputation and thus sales of its cars.
- It won’t make enough money in the good times to offset #1 and #2.
While the first two are still true, today’s focus is on the third. Tesla isn’t even making enough money to justify the first two risks. It’s actually losing money. A lot of money.
The Tesla Insurance Stack
Before going through the numbers, it’s important first to address what we can and can’t know about Tesla’s insurance ops.
Tesla writes insurance through three different entities – Tesla Property & Casualty (TPC), Tesla General Insurance (TGI), and Tesla Insurance Services (TIS). The latter is by far the largest. It’s also written through State National.
This means we have less info on the TIS piece, because State National doesn’t break out results for each program they write. Thus, we only know the total written premium, which is $387M. We also know this covers policies sold in California, Arizona, Illinois, and Ohio.
Outside those states, Tesla writes on either TPC or TGI. Since these entities only write for Tesla, we can examine them in greater detail.
TPC wrote $82M in Texas, Colorado, Maryland, Minnesota, and Utah. TGI wrote $28M in Virginia, Nevada, and Oregon.
If you add up the three entities, Tesla is writing $500M of premium. Over $50M is in Texas. We don’t know how much is in California, but given the distribution of Tesla automobiles and estimates of the other states, it’s safe to guess it’s over $300M.
Profitability
Since the bulk of the book is in TIS, we can’t look at loss ratios for this business. We can look at the TPC and TGI results as a guide though. Spoiler alert: they’re not good.
TPC had $51M of earned premium. It produced a 120 loss ratio!
TGI had nearly $22M of earned premium. It performed better…barely. The loss ratio was 119.
We can also get direct loss ratios by state. These will be somewhat better due to the lack of LAE attribution.
| State | Earned Premium | Loss Ratio |
| Colorado | $9.0M | 112% |
| Maryland | $8.6M | 119% |
| Minnesota | $1.5M | 123% |
| Nevada | $7.4M | 128% |
| Oregon | $4.1M | 116% |
| Texas | $29.6M | 116% |
| Virginia | $10.1M | 101% |
| Utah | $2.6M | 106% |
Amazingly, not one state is below 100%! Remember, this is loss ratio, not combined!
Busted Thesis
Recall, the Tesla thesis for entering auto insurance was they had better data on their drivers and could thus charge lower rates than the big insurers who didn’t understand Tesla drivers were safer.
Well, they did come through on the lower rates! Unfortunately, those rates weren’t adequate for the losses their drivers incurred. So much for the thesis about having better data.
Now, one might think Tesla would see these results and reconsider their push into insurance. If you think that, you obviously don’t know Elon.
No, Tesla plans to not only enter new states this year, they’re also going to start writing California on their own paper. Because what could be a worse idea than committing capital to a state that won’t let you raise rates on an underpriced product!
Now, you may want to give Tesla the benefit of the doubt and say they don’t have enough data yet to get the pricing right and it will improve over time.
The Path to Profitability
While that’s possible, think about what has to go right to pull that off. First, they need to raise rates a minimum of 50% to get to reasonable loss ratios.
Given the bulk of the book is in California, how is that going to happen? One of the issues I warned about in the earlier piece was the concentration risk in California given Prop 103.
It is extremely difficult to remediate a troubled California book, as we’ve all been reminded of over the last two years.
Outside of CA, they can raise rates 50%, but how will that go over? Not only will retention fall off a cliff, Tesla owners might get pretty upset.
One of the things I warned about in the earlier piece (see point #2 above) was reputational risk to Tesla from a poorly run insurance business.
If someone doesn’t buy another Tesla because they’re mad you raised their insurance premium $500, you have done a lot of damage to the brand!
So, if Tesla is ever going to be profitable, they first need to get smaller and risk their brand with existing customers.
A Better Future?
Let’s take an optimistic view though. Tesla grinds through the next couple years of re-pricing the book. They come out of it smarter and with only limited brand damage.
What happens next? Can they ultimately succeed? It’s going to be very challenging.
They haven’t proven they have a pricing advantage, so what reason is there for a Tesla owner to leave State Farm for Tesla Insurance?
The brand damage risk will always be there, unless they can find a way to save people money, not have dramatic swings in renewal pricing, and provide better service after a claim (which there is no evidence of yet).
The hurdle is just too high. Returning to the main points from the original thesis, there is not enough profit to be made for the risks of damaging the brand and the compounding of operational risks.
Tesla should exit the insurance business immediately.
Another constraining factor for Tesla is that Tesla insurance makes sense covering Teslas, but not any other car make (unless they can successfully white-label the algo which is not at all likely). Being a successful carrier long-term requires having a unique (and well-defended) niche providing outsized profits or average profits and scale. Tesla will never have either.
Thank you. Question: why would State National take on the balance sheet risk if they know Tsla stuff won’t be profitable?
Things may have changed, but I know when I worked with State National they were a strict front; just rent-a-paper. State National would paper the business, but take 0 risk (not even credit risk) via their tri-partite agreement between the producing entity and the ultimate risk-bearer.
State National very likely has no skin in this game, other than the fee they collect for papering the business.
While I can’t speak to the Tesla program specifically, I can tell you that State (and all the other fronts) have gotten looser over the year and now take not just credit risk, but also will take small participations in programs.
It’s not every deal, but it’s not uncommon. Too much competition so terms loosened.
Brian – State National doesn’t take on any balance sheet risk other than that TSLA won’t compensate them for losses. They have a 100% quota share back to a Tesla owned platform.
Back when this was launched one of the perceived opportunities for TSLA was managing the loss cost by being able to more firmly encourage the use of their own service and parts. At some level of loss cost part of the margin lost on the tax advantaged insurance entity is margin gained by Tesla’s service centers. That loss ratio probably isn’t +100% but its no where near as low as what a regular insurer would need either.
The other piece here is what is the expense ratio – we know its at least 6% on the State Nat fronted business, but how they are managing the marketing and LAE may figure as much in their long-term commitment here as anything else.
To the positive – I’ve been surprised they could generate as much as $500M in premium. I’d have never guessed a number quite that high for basically a one model insurance company.