Hacker News new | past | comments | ask | show | jobs | submit login
Self-Driving Cars Might Kill Auto Insurance as We Know It (bloomberg.com)
64 points by rmason 30 days ago | hide | past | web | favorite | 76 comments

This is some confusion about this, most of our insurance premiums go to paying claims. Most insurance companies operate with a slight "underwriting loss" or slight "underwriting profit", but usually quite close to even. The companies then make money by investing the "float".

So premiums going down by 90% is an implied 90% reduction of harm from cars. That's both property damage and human injury.

Warren Buffett: "...so vigorous indeed that it sometimes causes the P/C industry as a whole to operate at a significant underwriting loss. This loss, in effect, is what the industry pays to hold its float". [1]

[1] https://www.businessinsider.com/warren-buffett-insurance-flo... (with a nice explanation).

GEICO and Berkshire Hathaway are a unique beast in the industry. Their ability to gainfully employee capital is somewhat unmatched. Historically many companies managed to eek out an underwriting profit but distracted driving and bodily injury changed the game for many auto underwriters over the years following the iPhone's debut. Just in the last year or two or so years have some companies returned to modest profitability in this line. If you are not Warren Buffett your need to generate an underwriting profit is more or less essential. Weak returns on low risk investments vehicles which have performed worse in a post crash world make it even harder for insurance companies in general and make underwriting profit even more essential.

If you are a multi-line writer at this point you are well aware of the impact that autonomous, vehicle services and mobility services will have on auto. At this point its slightly better than a loss leader for other more profitable lines.

There is no confusion, 90% reduction in premiums directly leads to a 90% reduction in “float.”

Wouldn't the insurance companies keep any float they'd received up until the sharp decline in coverage costs?

Yes, "float" is on the asset side of the insurer's balance sheet. As coverage costs decline, the liability side of the balance sheet will shrink, causing the insurance company to be overcapitalized. Shareholders will then demand that the excess capital be used to pay for share buybacks and dividends (thereby returning the excess capital to investors so that it can be more efficiently allocated).

The best insurer is the one with the most data and ability to analyse it. Nobody can beat the self driving car operator, they need that information anyhow to improve the capabilities of the car.

Interest alignment is also with the self driving company, they have any incentive to improve safety of the car. They are simply liable so they act as an insurer and just add that to the cost of the self driving module.

For the buyer this should remove most need of insurance. Even if there is some "manual mode" the obvious provider would also be the self driving operator. They could track driving behaviour and adjust the rates, or offer some limitations for accident prevention.

So the insurance itself is still there, but it will be less important and shift to the self driving operators. Even if you own a car, you would pay some fee for the self driving module for updates and liability in one package.

I worked for insurance companies. Car insurance is a huge business, but not a very profitable or even sold at a loss. It is often used to get new customers to offer them other more lucrative products. That going away is still a huge shift and those companies are very old and slow.

With premiums plummeting then laws about liability can rise.

For example in California by law you only need $30k coverage. In Alberta it's $200k (most have $1 Million). This is really surprising given in California an ambulance can cost that much, whereas healthcare is nearly free in alberta ($250 ambulance iirc).

30k mandatory coverage? Is that a joke? EU has nearly free healthcare everywhere so car insurance almost never pays for any immediate treatment in accidents, and the legally required minimum is 5 million euro in 3rd party liability. On my own insurance I pay about £450 a year and my 3rd party liability is 20 million euro(and I didn't even select that as an option, it's just what it is).

Just speculating, but could it be since there's very limited ability to file for personal bankruptcy in most European countries?

Even if that was the case(I don't believe it is - there's nothing difficult about filling for bankrupcy it's just that the consequences can be rather severe - you can't run a company or get any kind of loan for years and years afterwards), why would car insurance companies go above and beyond what is required by them? My insurance is with Aviva, and I know as a fact that they are only required by law to offer 5 million euro coverage. Yet my policy clearly states 20 million euro 3rd party coverage. Why? I didn't ask for it at any point, and it still seems far beyond anything that I could ever cause in an accident. And it's certainly not like they advertise this fact anywhere, so it couldn't even be a marketing tactic(hey come with us, we offer more coverage than any other company!). It's just their bog standard policy as far as I can tell.

They will also kill government revenue raising from speed cameras. A billion dollar figure here in Australia.

Not unless there is legislation preventing cars from being told to drive above the limit.

I set my adaptive cruise control over the limit so I catch up to the car in front of me, for example.

Your adaptive cruise doesn't "know" what the speed limit is. Self-driving cars have this information, whether from a database or by reading a sign. It remains to be seen how much enforcement the software will perform.

My car does. Whether that's connected to the cruise control system is an open question.

I believe the tesla adaptive cruise control knows it in some cases. It will go max 5 mph over the limit on 2-lane farm roads (adaptive cruise control + self steer).

Unlikely. Car owners will still need liability insurance, lenders will require comprehensive, collision, and uninsured motorist coverage to protect their collateral. These coverages should in theory get less expensive if claims drop, but profit margin on the policies doesn't necessarily change.

> Car owners will still need liability insurance

Legal liability will shift to the manufacturers, therefore individuals will not require insurance. Insurance policies will be purchased by the manufacturers and will cover the entire fleet under one policy.

> coverages should in theory get less expensive if claims drop

This is certainly the largest factor, as insured losses are expected to drop by 90%. A secondary factor will be that car manufacturers will be able to negotiate wholesale “group policies” on large fleets and get a better rate. Or they could self-insure and cut out the insurance industry completely.

> but profit margin on policies doesn’t necessarily change

I have already noted the profit margin pressure arising due to wholesale fleet insurance / self-insurance. But the more important point is that profit margin is not the only relevant measure for an industry. The total amount of revenue is equally important. If profit margins stay the same but revenues decrease, then there is a reduction in net income. Note that prices are expected to fall 90%, which directly impacts revenue.

> Legal liability will shift to the manufacturers

No, it probably won't. Owners will probably remain liable in the first instance, for maintaining th vehicle in a safe operating condition; manufacturers (and the whole chain of commerce) will be liable, as they are now, for defects that exist at the time of sale, and will be (as they are now) attractive deep-pockets defendants.

But, for that reason, they will probably get very good at playing off failures (since drivers aren't in the picture to blame) as maintenance failures rather manufacturing defects.

The concept of "Owners" is an interesting one in the context of self-driving, manufacturer-insured vehicles. I'd imagine that self-diagnostics would have to get very very good to the point of not starting when a single sensor is unavailable - nobody would want to ride in "their" car if it has a chance of causing a crash and them being legally liable despite not driving it.

But, for that reason, they will probably get very good at playing off failures (since drivers aren't in the picture to blame) as maintenance failures rather manufacturing defects.

Manufacturers already do that now...Ford was doing that back in the Pinto days.

The change to self-driving cars will make maintenance (and maintenance records) more important for drivers for liability purposes, but will also make it more difficult for the manufacturers to avoid liability.

I was recently wondering why new cars don't come with factory fitted dashcams. Perhaps a law should mandate their inclusion, along with other data logging mechanisms.

Tesla does. Stick a usb stick in my model 3 and it uses the storage to record off the cameras. They just added a mode to record video when the car is “threatened” while parked.

> The change to self-driving cars will make maintenance (and maintenance records) more important for drivers for liability purposes

For owners. Until we recognize AIs as legal persons, self-driving car drivers don't have liability concerns.

No, it probably will. In the same way that you are not liabile for any accidents if you are a passenger in a taxi. It will take some time for legal scholars to work out all the details (e.g. gross negligence, etc), but it is very obvious that passengers in fully autonomous vehicles cannot be legally liable to the same extent that drivers are liable today.

> In the same way that you are not liabile for any accidents if you are a passenger in a taxi.

A passenger is neither an owner, an operator, nor a manufacturer of a taxi.

Owners, operators, and manufacturers tend to be liable; passengers not under normal circumstances. If I get hit by a bus (to escape the taxi situation where the operator is usually the owner), the bus driver, bus line, and bus manufacturer can all be liable. Replacing the operator with a piece of equipment manufactured by the manufacturer and maintained by the owner takes the driver out of the equation, but there is no logical reason why it would remove liability from the owner.

> Legal liability will shift to the manufacturers

Why? If you're operating a vehicle with known and manufacturer specified risks, thats on you.

It's basic product liability law. The liability is actually on both the owner and the manufacturer (and the dealer, and anyone else involved in that chain of transactions...). And as the deeper pocket, realistically the liability is all on the manufacturer.

You're not operating a vehicle. You're a passenger in a vehicle operated by waymo.

I thought we were talking about your own vehicles. If you own a self driving vehicle, and it crashes because you neglected maintenance then that's on the owner not manufacturer, just like it is with existing cars.

Sure, but what if I had done all the maintenance I was supposed to and it crashes. Should that still be on me or on Waymo?

I wonder self driving car manufacturers are going to try to get around this making the required maintenance and 'pre-flight' check list so onerous that basically nobody is completely in compliance and thus they're never responsible.

Yes, if you drive around and suddenly your spring coil breaks making you swerve and hit someone else, that's still on you, even if you have followed the maintenance booklet precisely. You can't sue the manufacturer for that, they will just say the part has worn out and you were supposed to spot that(somehow, even though the law requires an annual inspection at best).

if you drive around and suddenly your spring coil breaks making you swerve and hit someone else, that's still on you

Yes, but is it on me, the drive, or me, the car owner? Or if someone else is driving my car who's fault is it then? In the self driving scenario couldn't you make the argument that while it's my car, I'm only the passenger and Waymo is in fact the driver.

But you are the owner of the vehicle, right? Then you are responsible for its technical state, even if someone else, be it a person or a machine, is driving.

Like.....if you let your friend drive, and the tyre bursts and you end up hitting another car for example - no one will be at fault(most likely), but your insurance will still end up paying for it. I don't see why it would be any different with AI driving the car, in essence it simply doesn't matter. An object belonging to you has damaged an object belonging to someone else - therefore your insurance has to pay for it.

But you are the owner of the vehicle, right?

Then I won't own the vehicle, I'll rent it and leave the maintenance up to the rental company.

Who would pay to own a car that sits around parked >90% of the time anyway? A rental company could have that car out making deliveries and ride share pickups until it's time to pick me up at the end of the work day.

Then sure, absolutely, but there will always be people who will want to own the car not rent it for a plethora of different reasons that perhaps do not apply to you. And the argument here is that those people who chose to own self-driving cars will still have to insure them - the fact that the car drives itself doesn't change the fact that you could be liable for any damages that happen when it's on the road.

And as an aside - how is this different than a taxi then, at that point? If that model already works for you, then uber fills that niche already. Unless there is some assumption that somehow it would end up being cheaper than uber? I can bet it would be cheaper for me to take ubers to work than own my car, and yet I'd prefer to have my own person vehicle and not deal with shared vehicles.

uber fills that niche already.

Uber doesn't fill that niche. I'm talking about a long-term rental whereby the car is guaranteed to be available to me during specific time windows. Uber is only for one-off trips, not a long-term replacement for my commuter car.

I feel like that's a distinction without a difference. If I call an uber, it will arrive in 5 minutes max and take me to where I need to go. I can call one in the morning to go to work, and another one after work to get home. It could 100% replace my commuter car that I have right now.

Then why haven't you replaced your commuter car with Uber?

Because I like having my personal car, where I hop in and the seats, temperature and music are set to what I like, there's no weird smells and I have privacy and comfort. Besides, I think that about 60% of taxi drivers should never be in charge of any motor vehicle and are completely incompetent at driving, I don't feel safe at all. But that's beside the point and will be fixed by autonomous cars. But being in a vehicle shared by others will not, and I have no interest in doing that, just like I have no interest in taking a bus.

But yes, I have just done the calculation and it would in fact be cheaper for me to uber to work than to own my car. Yet I still prefer to have my own car that yes, does absolutely nothing for 95% of the month.

I still think fully autonomous cars would let you have your cake and eat it too. Maybe you don't want smelly passengers in your car or to have it unavailable at certain times of the day or week. That could be done. Maybe the rental car is available to you during the day but moonlights as a driver for Amazon packages?

It could all be spelled out in a service level agreement.

Yeah absolutely, that sounds completely fine to me.

> Legal liability will shift to the manufacturers

Nah.. Manufacturers will even invest heavily in lobby efforts for this to never happen. Even if consumers will likely not own the car as they currently do (see the growing trend that you're not actually owning the car but you're paying for the license to use it indefinitely) they will be responsible for keeping it running will be the de-facto 'operators' regardless if the car is driving itself. The car will probably not store itself, not maintain/clean itself, will not keep it's radars/cameras active and not blocked by dirt/snow/etc.. Call me a pessimist but the consumers will likely get shafted.

This is a battle they lost decades ago. Product liability laws have been on our books for more than a century, and the general gist is that everyone from the owner to the manufacturer is jointly and severally liable to various degrees depending on the circumstances of the failure.

A failure to do poor maintenance shifts the liability almost entirely to the owner, but a defect puts liability almost entirely on the manufacturer. (But note: "jointly and severally liable" means that they're all on the hook with respect to the plaintiff's award. The assignment of liability is how the plaintiffs ultimately settle the bill amongst themselves.)

> This is a battle they lost decades ago. Product liability laws have been on our books for more than a century, and the general gist is that everyone from the owner to the manufacturer is jointly and severally liable to various degrees depending on the circumstances of the failure.

If the battle was lost, where's Tesla wrt to the deaths caused by autopilot? Or was the manufacturer's disclaimer enough to blame the dead operators?

> Call me a pessimist but the consumers will likely get shafted.

You're a pessimist!

Actually I don't see that there would be much difference. The car will perform at some level of skill and have an associated risk, lets say that amounts to $50/mo. The customer could pay the manufacturer to pay the insurance company, or the customer could pay directly -- same thing?

The end game is that almost no one will own a car, and most people will use fully autonomous taxis for transportation. If and when fully autonomous vehicles become a reality, auto insurance will die along with car ownership and parking lots.

I see little reason to suppose that that will be the end game. Those for whom the total cost of ownership of a car are lower than the cost of taxis will continue to own a car.

What fraction of people that currently own cars will that be? Probably depends on where you live. Taxi will probably never be the popular choice in, say, South Dakota. The economics just don't work out.

The cost of autonomous cars, especially in terms of maintenance, is likely to be higher. And autonomous taxis will be cheaper to operate because drivers are a very expensive part of them. The number of people for whom owning a car makes sense will definitely go down when prices on cars rise and prices on taxis and public transportation drop.

This is ignoring the fact that having an unproductive self driving car sitting in your garage becomes a significant opportunity cost itself.

I really doubt that.

Shared autonomous cars would end up being treated even worse than rental cars. People would smoke in them, vomit in them, damage the upholstery, steal the radios, let their pets go to the bathroom, spill all kinds of stuff in them on the way home from Home Depot, and so forth.

Anyone who can afford it and has a place to park it will want to have their own self-driving car to avoid having to put up with the shared ones.

Ownership will also be more convenient, e.g. you can leave your stuff in your car when you park it. Your car is reserved for you to use whenever you want, so you don't need to wait for a shared car to become available.

Stick a video camera inside the car. If an ML algorithm detects anything too far outside of the norm, send the video to a human. If the human notices damage, the riders are banned from the system. Banned rider lists could be shared between autonomous taxi providers. End result: no one will fuck with the cars.

Except that we are humans not robots. You never picked up a friend after a night out who got sick in your car? Or had a baby that decided to smear poop all over the seats? Spilled coffee in the car? Know someone who likes to smoke in the car? That stuff happens, and if there was even a slight risk that it would get you banned from convenient transportation then yeah, people will buy their own so they don't need to worry about this. Pretty much everyone who has kids has to keep extra stuff in the car just in case - ergo, they will always go with personal vehicles. Otherwise, we'd see far more people switching to taxis than driving their own cars, especially compared to new cars, as monthly taxi cost can easily be lower than car/insurance/tax/fuel payments on a new vehicle, and yet people keep buying them.

It would be very simple for manufacturers of autonomous cars to obviate the question of "who is liable" for a collision by simply loading insurance into the price of the vehicle.

Extremely easy to price into a lease; less obvious but still eminently feasible with a sale.

I am going to go with, not until they get all non self driving cars off the road. Until then during the transition it will become a lawsuit paradise until the regulations are settled as any crash involving a self driving car with a person driven car will be challenged with boasts of class action suits and more.

a large portion of my current insurance is for uninsured motorist and under insured though my agent says the former is by far the biggest issue. considering law enforcement isn't making much head way there and true self driving isn't going to happen soon except under theme park ride conditions I don't think insurance is going anywhere.

I assume I'll have to buy insurance to go into 'manual' mode for a few minutes at a time, the price fluctuating according to the time, where I'm driving, and a possible breathalyzer-style test ;)

House insurance still exists, despite the fact that the rate of collisions caused by human error is zero. I’m sure premiums would change considerably, but “kill”?

…houses are insured for different things than cars are?

Home insurance is considerably less expensive than auto insurance today. If auto insurance prices fell to the level of home insurance, then it does effectively kill the industry as we know it. It’s not a very good comparison anyway, as home insurance covers a variety of perils which are often caused by human error (e.g. fire), as well as natural perils that are not really applicable to auto insurance (e.g. flooding due to plumbing issues, or roof damage due to wind storms).

Umm... premiums depend entirely on who you are, where you live, and what you're insuring... I pay more than twice as much for home owners insurance ($125/month) than I do for car insurance ($60/month).

Cars flood, catch fire, and get damaged by storms too.

home insurance covers personal liability of the insured and the premises - and the rate of "collisions" is very far from 0.

Cars will still crash. We have 2 options - build the cost of insurance into the price of the car, or let owners buy it themselves.

IMO Id rather have the ability to buy my own insurance and have various choices of coverage. Nothing needs to change.

I think what will change is... take a South Floridian car insurance car payment. You can pay $200/$300mo for 60 months and never need it, because based on statistics, everybody else in your area is getting into accidents with uninsured motorists.

Less accidents happening around you, insurance claims write fewer checks for claims, your payment (which is basically a savings account contributing to "when you might need it" in my eyes) should be lower.

I'd pay a small premium if it's built into the a single ongoing cost of a car. I know I'm not alone - Insurance industry is one of the least trusted.

This may be bi-modal - many many folks would agree with you, but enough might agree with me for someone enter the market - potentially disrupting it.

How am I liable for an accident caused by my self driving car, though?

There are other things your insurance covers, like an uninsured driver hitting your car, theft, a tree falling on your car, etc. These things will still happen to self driving cars.

> How am I liable for an accident caused by my self driving car, though?

Because you failed to maintain the vehicle in a safely operable condition. Now, if s manufacturers defect contributed to that failure, the manufacturer may also be liable, but that doesn't mean you aren't liable, and the manufacturer is going to claim and try to prove that that isn't the case.

In practice that sounds impossible. Am I supposed to audit the code? I probably don't even get to see the code, and even if I did we don't have any even theoretical way to audit the neural network parts.

But you can audit tires, physical damage to sensors, cameras, etc so if an accident were to be caused by failure to brake caused by rain and bald tires, that is entirely your fault. ABS will only take you so far.

> In practice that sounds impossible. Am I supposed to audit the code?

I know HN has a software bias, but a self-driving car is not a blob of disembodied code, and the code that is present relies on the state of hardware for proper operation.

The self-driving car manufacturers and downstream vendors themselves will need tremendous coverage when any accident can suddenly pin at least partial liability on them. That cost in turn will be passed down to consumers.

I completely disagree - seems like pricing insurance for self driving vehicles competitively would be a great way to ensure more people transition to self driving vehicles rather than conventional vehicles.

I agree with you. Self driving cars will become the norm the minute it's cheaper to insure the automated car instead of the human being

nit: Once the price difference is greater than the marginal cost of self driving equipment

Yet another 'might' article....

This article has it backwards: payouts will drop, so premiums will drop (less drastically if they can get away with it). In fact once autonomous vehicles really get working it will likely become very difficult to get permission to drive a vehicle instead of having it drive itself.

The insurance companies were instrumental in the development of things like fire codes, bank safes and the like; they hate paying out claims because their business model is to hang onto the premiums and invest them.

Yup - driving your car will go the way of the diving board in a pool (not illegal, but outrageously expensive to insure)

Applications are open for YC Summer 2019

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact