Driverless cars + Uber –> Death of Car Insurance ?
As this became a long post, a quick management summary upfront:
The case for 100% self driving cars without accidents is not so clear for me
1. Based on current facts, the Google car doesn’t seem a lot better than human drivers
2. From other areas (Airplanes, chess) we can learn that a human-machine combination is often better than a “machine” alone
3. Driving cars is also an emotional experience, many people might not fully sacrifice this
4. Some innovations take longer than one thinks, especially if they take away freedom from consumers
5. A gradual decrease of claims could actually be positive for car insurers over an extended period of time
Additionally, I don’t see a combination of driverless cars with a service like Uber replacing private cars anytime soon. There are a lot of practical issues with renting out private cars to complete strangers. However, taxi driver might not be a job with a big future either.
So from my perspective, as shareholder of a car insurer like Admiral there is no reason to panic, however for traditional insurers this might be one more nail in their coffin.
On the weekend, I linked to a post from the brand new “Bank of England” blog.
The major argument was the following: Self driving cars will in the future radically decrease or even eliminate accidents and therefore damages. With no accidents and a driverless car, the current business model of selling car insurance to drivers basically losses its reason to exist.
In a second step one could even build on this first assumption: If we really get well functioning self driving cars, then a combination with a “Uber-like” service will greatly reduce the number of cars in total. Imagine a car sharing app where you press a button and 2 minutes later a nice shny driverless car appears in front of your door to take you wherever you want for a price cheaper than even your subway ticket.
Clearly such a scenario would not be good for car insurance (and neither for car manufacturers). In general, I do think it makes a lot of sense to think about what can make a company obsolete. There are many examples of once highly succesfull companies which got killed by technological progress, the most famous being of course Kodak. The “driverless car kills car insurance” argument by the way is not that new, if you google around you find posts already more than 2 years ago looking at this problem:
The doomsday scenario is clear for the roughly $200B in personal and commercial auto insurance premiums written each year in the US. Insurance premiums are a direct function of the frequency and severity of accidents. In a world of driverless cars, where accidents are significantly curtailed, most of those premiums will go away. Sure, some car insurance will be needed, but the market might be reduced by 75% or more. Insurers make their profits on the float from their premium income, so plunging premiums spells doom for many insurers.
Let’s start with the “driverless car” aspect.
1. Fact checking
This is how the BoE post starts:
In 2020 Google plans to launch a self-driving car which has already driven nearly one million miles without causing an accident; it doesn’t get tired and irritable, swerve into lamp posts or require a driving test.
From that “data” they automatically conclude that the Google car is better than a human driver. Well, where are the facts that are proving this ? This is an article from the LA Times which looks a little bit behind Google’s “advertisement” and this is the link to the original post from Google. This was the quote which got most people excited:
If you spend enough time on the road, accidents will happen whether you’re in a car or a self-driving car. Over the 6 years since we started the project, we’ve been involved in 11 minor accidents (light damage, no injuries) during those 1.7 million miles of autonomous and manual driving with our safety drivers behind the wheel, and not once was the self-driving car the cause of the accident.
In the LA Times article they however pointed out what Google didn’t say for instance:
The post indicated that the cars were driving themselves for nearly 1 million of the 1.7 million miles they were on the road, which is close to 60% of the miles traveled. But there’s no telling whether the 11 accidents occurred when the car was in self-driving mode or being manually operated.
Nor did Urmson say whether the collisions tipped Google off to a weakness in the cars’ awareness of the world around them, leading to a software update that should help cut down on future collisions. There was no indication whether the number of accidents per mile traveled increased or decreased over time. We don’t know whether the human driver inside Google’s cars ever had to take control to avert an accident. And Urmson provided no clues whether the accidents were correlated with particular traffic conditions — say, for example, urban rush hours.
Another interesting aspect: The Google cars are actually not hat much better than human drivers, again from the LA Times article:
By the way, 11 accidents in 1.7 million miles, or 6.4 per million miles driven, seems right in line with the typical American driver’s experience. Using data from the National Highway Transportation Safety Administration from 2008-2009, the AAA Foundation for Traffic Safety estimated that U.S. drivers got into wrecks 4.2 times per million miles traveled on average. The AAA figure understates the average, though, because it’s based only on the accidents reported to police, which represent a little less than half of all collisions. Factoring in unreported collisions, the average increases to 9.3 per million miles.
However a headline called “Google self driving cars slightly better than human drivers” doesn’t sound that eye-catching anymore I would suggest.
So with regard to fact checking I would argue: From what we know, the current version of the self driving car doesn’t seem to be much better than the typical US driver, although we don’t really know for sure as Google doesn’t share its data. So yes, maybe in the future the algorithms get better but at the moment they are far away from eliminating accidents.
2. Airplane autopilots and chess programs
Have you ever wondered why we still have human pilots on airplanes although autopilots for airplanes have been introduced as early as 1930 ? Well, some might argue it would be safer without them but you can be sure if this would be that case, airlines would have kicked out pilots many years ago as autopilots don’t go on strike. Well, it turns out that humans still seem to be better prepared to handle unexpected situations. And flying an Airplane through quasi “empty” air is a lot less complex than for instance navigating something like this:
Even in a pretty regulated game like chess it is quite interesting that although machines beat humans, the combination human+ machine beats the machine by a wide margin.
Going back to the Google car: Whenever I was driving myself in the US and especially in California, I was always amazed how easy and relaxed driving was. Wide and straight roads, big parking places without need to reverse, drivers actually drive according to the rules, nice weather etc. etc. Compare this with driving in Southern Italy or even Asia. Chaos everywhere, Scooters zooming around, people plainly ignoring rules, huge potholes in the streets, chaotic street design, construction sites & road work etc. Or in my home town Munich where we frequently have snow on the street or very icy side streets etc.
Or how will the Google car handle “French parking” ?
My argument here is that it will most likely take a long time until computers can handle very complex, chaotic traffic like a human can and most likely a combination of human and “assistance” will be better than the pure machine when it comes to complex situations.
Combined with point 1) (lack of data from Google) I do think we are still some years ago for a fully self driven car which works under ANY circumstance and is REALLY better than a human driver or a human assisted driver. And as long as you need the human, the insurance needs to be personalized
3. Humans will continue to want to drive themselves
Living in Munich, the headquarter of BMW with their famous slogan “Freude am fahren” you can easily see that many people don’t buy their car just to get from A to B. They buy their expensive cars because the car companies somehow made driving cars an emotional experience. Although I am not a big car buff myself, I do enjoy driving cars myself. Not if I need to go 8 hours on the Autobahn, but if I go out on the week-end to the mountains or visiting a nearby city, I do actually enjoy driving myself.
I do believe that there always be a relatively large group of people who want to drive themselves. Or take my “better half” as an example. She often is late for appointments. Would she be willing to sit in a driverless cr which goes exactly as fast as the speed limit or would she “over ride” the auto pilot to go faster ?
4. Innovations sometimes take much longer than people think
Consultants these days often repeat that innovations are implemented faster and faster. It took for instance 40 years or so until most people had a land line phone and it took only 10 years or so until the mobile phone “penetrated” the same amount of people. On the other hand, people still do read newspapers and magazines (maybe a little less) despite being able to access everything only.
My unproven theory is the following: innovation do get implemented faster these days if they add something to customer’s life, especially more freedom. For instance the mobile phone added access to communication everywhere without taking away anything (you could keep your landline phone in parallel, no problem).
A self driving car adds clearly something: You can use your time better than just looking at the road. On the other, it takes away some of your freedom, mainly having full control over your vehicle. Especially in Europe, there is also a strong tendency against being “controlled by the machine”. Add on top of this the typical car replacement cycle of 8-10 years then one could also construct scenarios where the penetration of fully self driving cars takes much longer than expected.
For car insurance however this means that as long as you are in principal able to override the autopilot, you will need to have personal insurance. I personally think that the 100% autopilot car without individual insurance will remain a small minority for quite some period of time.
Another aspect of why this might take (a lot) longer could be that if they are smart, car companies will not let Google into their cars. Having seen how Android takes away the meat for mobile phones, most car manufacturers will do anything to keep Google out. I think they will gradually add assistance systems like they do already now but not go for full autopilot anytime soon, especailly not the premium brands.
5. What is the impact on car insurance companies ?
I do agree that a sudden dramatic drop in the frequency of claims will over the medium term be bad for car insurars in general. In the short term it will be positive as car insurance premiums are much more “sticky” than claims ratios. This is not theory but based on historical evidence. For instance, the most profitable car insurance markets in the last 3-5 years were: Not China or Germany but Italy, Spain and Greece with Combined Ratios of around 80%. The reason is simple: The frequency of claims correlates directly with thenumber of kilometers or miles driven. Less money for gasoline means lower mileage and lower claims. Usually it takes 2-3 years until insurance premiums catch up with claims. Often this creates a problem to the other side of the cycle as well. This is for example the case in Ireland, where now in the recovey miles driven increase but insurance premiums ar lagging which creates problems for the local players, such as IBD at which I looked a while ago.
A slow retreat of claims however might be very beneficial for quite some time if this developement is gradual as the claims go down earlkier than the premiums. Within the insureance industry, clealry players with a higher cost base suffer first. For a cost leader like Amdiral one could even create a case where they gain further market share because “traditional” insurers will need to give up pretty early in the process.
In the very long run however lower claims and lower premiums are clearly not positive as the overall cake is getting smaller and the business will become even tougher. Insurers could try to counter this decline with other products. A practical example what actually can be seen already are Extended Waranty covers, where the insurer covers a time period after the original manufacturer guarantee against unforeseen repairs etc. This could actually evolve into a potential extra business line as repairs get more expensive all the time.
One should also not forget that “Collision Damage” insurance is only part of the insurance premium for car insurers. You still want to be insured against theft, hail storms, scratches etc. which is a considerable part of the insurance premium.
I do think that for this and other reasons, “traditional” insurance companies with an expensive sales force are already deep in trouble and will be hit hardest and first by those potential developments. So my advice would be to stay clear of cheap traditional insurers with a large share of car insurance anyway.
So now let’s move on to the “Uber aspect”, more specifically (and more dramatically) the question if “Car sharing 2.0” would actually lead to a lower amount of cars actually being bought and owned by people than before.
For anyone who uses care sharing now, there are some obvious disadvantages. First, you need to find the car first. Usually, in contrast to advertisements, car sharing cars are often a very good walk away which is especially inconvenient if you have something heavy to carry. Secondly, often you have to leave the car inside a certain area. In Munich for example where I live, the allowed area is still two subway stops away from where I live.
Being able to actually call a car-sharing vehicle to your door steps within a very short period of time sounds actually quite attractive, it is just like Uber without the driver. The argument goes that cars are used only 20% of the time, during the rest, car sharing would greatly improve efficiency and decrease cost for everyone. If people prefer this to owning a car then both, car insurers and car manufacturers would be in dep trouble as a lower amount of cars automatically reduces business and much quicker than the case with declining claims.
For this to work (i.e. having a lower amount of cars in total), two things have to happen:
a) many people have to accept not owning their own car
b) many people wanting to share their car with complete strangers
For the first point, we again enter “emotional territory”. For some people cars are just a mean to go from A to B, for many other people cars are something different. They are either a status symbol (Porsche, Mercedes Benz etc.) or lifestyle assesoire (Mini, Smart) or just an expensive toy. Practically I am also not so sure if this would work for the typical commuter as during rush hour , most likely the rate of “idle cars” is a lot lower.
I would also be interested how a 2 week vacation trip from Munich to Tuscany would work, I assume once you talk about greater distances and longer periods of time, car sharing gets more difficult and expensive. also if you have kids for instance, it is quite burdensome to have to remove and install special children’s seats each time you want to go somewhere. In my hometown Munich, I know already many people who don’t own a car because the public transport system is so good or they go by bike. For them such a service would simply replace a taxi for unusual destinations or very late at night.
But also the second aspect of this concept is problematic. If you just bought yourself a nice shiny BMW for 100k, do you want some complete strangers to ride around in this car ? If your car returns home in the evening, how do you handle dirt, scratches and other stuff in the car ? Do you put in video cameras and then charge the user for this ? I do see a kind of negative selection bias here. People with old cars who don’t care a lot of them (and maybe do not have the money) might be much more willing to rent out their cars than owners with nice cars who are emotionally attached to them. So this kind of service then could quickly become a “call a wreck” kind of service.
What if a car rental company for instance would offer dedicated cars for such a service ? This would be a possibility, but then I would doubt that the overall number of cars would actually decrease. I think that this would rather replace the Taxi and/or rental car fleet.
All in all I don’t see a large penetration especially of people willing to share their car with strangers.
Self driving cars either stand alone or in combination with a service like Uber will clearly have an impact on cars and car insurance. As I tried to explain enough, in my opinion this will take quite some time and if for instance airplanes are a benchmark, there will always be a human aspect in driving.
I do not think that car insurance will be obsolete in the 10-20 years but there could be clearly a downward trend in the frequency of accidents mostly due to better and better assistance systems. So far however the available facts don’t prove that the Google car is as good as some people claim and the availability of date is not very good which in my opinion is something which has been overlooked in the discussion so far.
The impact on car insurance itself is not so clear in my opinion. If the decline is gradual, margins could actually go up for quite some time and “traditional” insurers will get hit first if overall premium declines. On the other hand there might be additional products like extended warranty etc. to compensate for this.
One should also not underestimate the human and emotional aspect of this. For people who don’t like to drive and only own a car because they use it as a pure means of transport, a driverless Uber service is a no-brainer. Everyone else will think hard of giving away freedom which comes with owning and self driving a car.
I also think that car sharing will not be the “killer app” some people think it is but let’s wait and see what actually happens.
In any case an exercise “what could kill your company” is always very helpful and should be done for any investment as there is always something which could kill a business model.