The future of the auto insurance industry

[WhoIsWarren]

A recent (1st November 2013) Grant's magazine brought into question the auto insurance industry and I thought it was worthy of a board discussion.

 

To summarise, self-driving cars or "autonomous vehicles" (AVs) are projected to be a mass-market item by 2022.  This would likely devastate the auto insurance market.  It potentially means "the end of auto insurance", as the the then-CEO of Progressive mused in 1998.

 

I've got to say that this made me sit up.  I've not heard any mention of this before; at least I don't think this is mainstream thinking.  In particular, I've never heard anyone question Buffett about it, about the implications for GEICO.  I think GEICO accounts for a growing one-third of Berkshire's insurance premiums.....running at a very profitable c.90% combined ratio (and probably lower given the high levels of investing for growth the company is undertaking).

 

Has anyone thought about this?  Perhaps you can direct me to some further reading on the topic.

 

Thanks

[JBird]

http://www.forbes.com/sites/chunkamui/2013/03/28/will-auto-insurers-survive-their-collision-with-driverless-cars-part-6/

[ItsAValueTrap]

The technology seems *far* from proven.  If the technology works, it will first be rolled out into easy situations such as roads with very little traffic (e.g. shuttle buses).

 

2- If you look at Buffett's portfolio, technology devastated the newspaper and encyclopedia (World Book) industries.  Overall, he still made a lot of money from those investments.

[Ham Hockers]

People insure things that are expensive. If I spend $50k buying a car I'm not leaving it uninsured, even if I don't think I'll get into an accident. Plus, there's always the chance a tree falls onto my parked car. Nothing AVs can do about that.

[Guest 50centdollars]

Also, what about hackers who can hack into a car's computer and start controlling the car. That wouldn't be cool.

[constructive]

People will still want the ability to override driving control. Accidents will still happen whether caused by man or machine. So states will still want to mandate insurance.

[StubbleJumper]

If it's a mass-market item by 2022, then it won't likely be dominant on the road until around 2037 as the fleet ages and older cars are scrapped.  So, we have about 10 more years of business as usual, followed by 15 years of transition as premiums drop drastically because the collision and upset portion will disappear.

 

So what do you end up with?  Perhaps in 2037 premiums in real terms might be one-quarter of the current aggregate amount (ie the comprehensive portion of existing policies), meaning that a pile of capital needs to be redeployed?

 

I think I'll worry about that in 10 years.

 

 

SJ

[JBird]

People insure things that are expensive. If I spend $50k buying a car I'm not leaving it uninsured, even if I don't think I'll get into an accident. Plus, there's always the chance a tree falls onto my parked car. Nothing AVs can do about that.

 

The idea isn't that auto insurance will disappear-- even if all vehicles are autonomous. It's that premiums will fall dramatically because the probability of an accident has fallen so much.

[writser]

People insure things that are expensive. If I spend $50k buying a car I'm not leaving it uninsured, even if I don't think I'll get into an accident. Plus, there's always the chance a tree falls onto my parked car. Nothing AVs can do about that.

 

The idea isn't that auto insurance will disappear-- even if all vehicles are autonomous. It's that premiums will fall dramatically because the probability of an accident has fallen so much.

 

But then payouts will fall dramatically as well. Ok, you have less float to play around with but it doesn't have to spell the end of the industry. Also, shipping and flying are largely automated (correct me if I'm wrong). What has been happening in the insurance sector over there?

[Hielko]

People insure things that are expensive. If I spend $50k buying a car I'm not leaving it uninsured, even if I don't think I'll get into an accident. Plus, there's always the chance a tree falls onto my parked car. Nothing AVs can do about that.

 

The idea isn't that auto insurance will disappear-- even if all vehicles are autonomous. It's that premiums will fall dramatically because the probability of an accident has fallen so much.

 

But then payouts will fall dramatically as well. Ok, you have less float to play around with but it doesn't have to spell the end of the industry. Also, shipping and flying are largely automated (correct me if I'm wrong). What has been happening in the insurance sector over there?

But a (significant) lower float is very bad for BRK. At the moment $1 in float is basically $1 in shareholder value: every single dollar that they need to repay is a loss.

[Ham Hockers]

The idea isn't that auto insurance will disappear-- even if all vehicles are autonomous. It's that premiums will fall dramatically because the probability of an accident has fallen so much.

 

From the article you posted: "Cripe agrees, though, that in the long term, 'car insurance goes away.'"

 

That doesn't sound like he's just worried about falling premiums.

[JBird]

I should've been more clear. I wasn't referring to the article.

 

Clearly, even if all vehicles are autonomous, the probability of an accident is not zero. Therefore insurance, however limited, is still necessary.

[wisdom]

Will the manufacturer's need to buy insurance - in case, of accidents caused by their self dirving cars.

 

The total premiums should drop due to lower accident rates.

 

This may leave families with more $s to spend on other things. There will be other opportunities.

[Ham Hockers]

But a (significant) lower float is very bad for BRK. At the moment $1 in float is basically $1 in shareholder value: every single dollar that they need to repay is a loss.

 

Unclear if this would be bad or good for BRK. Tighter margins could kill competitors. Aggregate industry & profitability could shrink but BRK could offset with larger market share.

[Guest ajc]

Isn't this one possible reason why Berkshire recently hired those AIG guys, among other things.

[Vish_ram]

People would still have to buy "autonomous vehicles" insurance.

 

The demand side of equation would change (go up) as more and more elderly who otherwise are not driving, would start using AV's. With the aging population that is living longer, you would see a massive demand for both cars and insurance.

 

 

[Peregrine]

Can technology completely eliminate the possibility of accidents? I highly doubt it. A level of human cognition will always be desirable. Whether it will make driving safer is an entirely different question. But I don't think people can handle not driving their own cars in one form or another anyway.

[JBird]

But I don't think people can handle not driving their own cars in one form or another anyway.

 

Why not?

[LC]

I think the second there is one accident involving self-driving cars, we will go back to buying insurance. And given human nature, there will most definitely be an accident...(and the obligatory lawsuit naming Google as a defendant).

[StubbleJumper]

I don't think anyone is really arguing that insurance will go away entirely.  Rather, the most expensive perils will drastically decline (ie, personal injury, followed by collision damage).  If collisions are virtually eliminated, what's left of car insurance policies will essentially be the comprehensive portion (vandalism, weather damage, break-in, theft, etc) and an extreme tail-risk of a freak collision.  Working by memory, for my car insurance the comprehensive premium is about 25% of my total premium.

 

So what would this mean for Geico?  Well if my rough estimate of a 75% decline in premiums is more-or-less correct and if the market size does not grow, that means that Geico's float in today's dollars, eventually needs to decline by perhaps 75% and about 75% of Geico's statutory capital can eventually be liberated and re-deployed to other uses.  Working again from memory, Geico's float is about ~18-ish billion, implying that maybe $14.5B would eventually disappear (by about year 2037).

 

I have trouble getting too upset about this prospect, as $14.5B of float is less than 5% of BRK intrinsic value, and the re-deployed capital will create some new intrinsic value to partially offset the loss of float.  And none of this even begins to occur for another 10 years and doesn't fully kick-in until 25 years down the road.

 

So there's a credible risk of less than 5% of intrinsic value disappearing some time in the next 25 years?  Who really cares?  That's just a bad west coast earthquake.

 

 

SJ

[Peregrine]

But I don't think people can handle not driving their own cars in one form or another anyway.

 

Why not?

 

Would people really feel safe or comfortable with a computer guiding their cars? Besides, even on planes and trains where everything is highly automated and human input is designed and forced to be minimal, humans still have to be at the controls.

[Peregrine]

I don't think anyone is really arguing that insurance will go away entirely.  Rather, the most expensive perils will drastically decline (ie, personal injury, followed by collision damage).  If collisions are virtually eliminated, what's left of car insurance policies will essentially be the comprehensive portion (vandalism, weather damage, break-in, theft, etc) and an extreme tail-risk of a freak collision.  Working by memory, for my car insurance the comprehensive premium is about 25% of my total premium.

 

 

 

A counter argument can be made that the most at-risk drivers will turn off their computers and drive however they please, while the least-risk drivers will keep their computers on (if it does improve safety). Overall the possibility of accidents may not change much at all.

[WhoIsWarren]

Thanks for all the replies.

 

JBird, good article....explains all of the main issues.  And as you said, it's not like there's never going to be an accident / theft / tree falling incident again, it's just that the incidence goes *way* down.  StubbleJumper.....a 75% reduction in premiums is probably as good an estimate as anyone will make; just to add, comprehensive insurance insures you for others crashing into you too, right....the probability of which will drop, no?

 

I know that this issue isn't going to bury Berkshire -- of course not.  That said, I previously had thought of GEICO as a growing source of premiums, earning a very healthy combined ratio. 

 

So just trying to work through some numbers.  According to the 2012 annual report, GEICO's float is $12bn.  Let's look at two scenarios.  In the first, the float grows by c.5% for the next 10 years and thereafter remains stable.  It earns a combined ratio of 92% (10 year average -- conservative I think) and generates a 5% investment return.  Using a 5% discount rate, this float is worth $44bn.

 

In the second scenario, the float grows at 5% for a few more years and declines gradually out to 2037 (random).  I've modelled the combined ratio declining over the years to 100% because of loss of scale and perhaps a dramatic shift in the route to market for auto insurance in the future (per the Forbes article).  Again using a 5% discount rate, the value of this float is around $15bn.

 

That's a material difference!

 

I'm not saying either scenario is likely or my estimates are very accurate, but I also wouldn't be so quick to say it's "just a bad west coast earthquake".

 

One thing to finish off -- from a 2013 interview with Buffett:

 

Becky Quick (CNBC): “If you could keep one company that Berkshire owns, either a wholly-owned subsidiary, or that Berkshire owns a common equity in, which one would you keep and why?”

 

Warren Buffett: “I would keep GEICO........"

 

Perhaps he's just being sentimental (I doubt it!), but he knows all about AVs I'm sure.  Yet he's still very optimistic about the future for GEICO!

[StubbleJumper]

 

 

So just trying to work through some numbers.  According to the 2012 annual report, GEICO's float is $12bn.  Let's look at two scenarios.  In the first, the float grows by c.5% for the next 10 years and thereafter remains stable.  It earns a combined ratio of 92% (10 year average -- conservative I think) and generates a 5% investment return.  Using a 5% discount rate, this float is worth $44bn.

 

In the second scenario, the float grows at 5% for a few more years and declines gradually out to 2037 (random).  I've modelled the combined ratio declining over the years to 100% because of loss of scale and perhaps a dramatic shift in the route to market for auto insurance in the future (per the Forbes article).  Again using a 5% discount rate, the value of this float is around $15bn.

 

That's a material difference!

 

 

Okay, and what have you assumed about the portion of GEICO's capital that will be liberated as the aggregate premiums decline?  If premiums go down by 75% then you can re-deploy 75% of GEICO's statutory capital to some other use.  At a minimum, you'd need to value that liberated capital on a one-for-one basis when calculating intrinsic value in your second scenario.  And it wouldn't be much of a stretch to argue that the intrinsic value of the re-deployed statutory capital could exceed 100% of its nominal value (assuming that there are decent investment opportunities that appear regularly over the next 25 years).

 

The second observation is that it might be a stretch to calculate an intrinsic value of the float over a 10-year timeframe based on a 92 CR and 5% annual growth in float.  In essence, there are a couple reasons why this is questionable: 1) underwriting quality tends to get undermined as you add market share which might imperil a 92 CR...and projecting 92 as a terminal CR is a bit scary; 2) market share starts to get very large as GEICO currently has about a 10% market share but 5% growth for 10 years would put it at around 16% market share.

 

So if you tone down the assumptions underlying your IV and you redeploy some statutory capital what is the impact?  A big earthquake?

 

 

SJ

[WhoIsWarren]

OK SJ, I getcha -- tone down the scaremongering  ;D ;D

 

Points well made, probably not a big deal.  And as Buffett is still so positive on GEICO, I'd say the truth is far closer to your not-a-big-deal scenario than my what-if scenarios.  ;)

 

As for the 92% CR, again you are right to question it's sustainability, but my understanding (prior to reading about AVs!) was that their direct model is so difficult to replicate, that GEICO is increasingly gaining scale and efficiency advantages over the competition, a virtuous circle.  Plus they've been investing so heavily in the last few years (for future growth) that if anything their reported CR is overstated.  Do you think that's not the case, or are you perhaps just being conservative?