Motor insurance: driverless cars and the new norm

In attendance:

Moderator: Christian Wuestner, deputy editor, Intelligent Insurer

What is happening in the motor insurance industry today?

Jonathan Dye: We’re facing unprecedented change in a lot of respects, not only from technology, ‘Internet of Things’, telematics or autonomous vehicles but also in relation to peoples’ attitudes and in particular the sharing economy. These are challenging but exciting times. As with any change, there’s opportunity for insurers, brokers and reinsurers to take the initiative and create solutions and propositions that our customers require in the future.

Crispin Moger: The reason Marmalade became involved 10 years ago is because nobody was really addressing this market, specifically the young driver motor market. After looking at it we thought, if we can find the real pain points and then find ways of mitigating those pain points then there could potentially be a really profitable business there, and that was our challenge.

A lot of people thought we were mad. With our learner driver product we went around pretty much every insurer and I think only one didn’t slam the door in our face, but we then showed that you can do it.

Do you have a competitive advantage compared to the big insurers because you’re taking a fresh look at the market?

Moger: Potentially in terms of customer insight, yes, but you still need the insurers. We need so much money for the motor side because of all the liabilities. Starting up from scratch is very difficult. Where we’ve done well is in understanding customers, understanding what drives young people, what they want and then how to make that product work for them.

Unipol has a big market share in Italy but you want to apply your expertise elsewhere in Europe. How do you assess the growth potential in other European markets?

Marco Sordoni: We started selling telematics products around 2003 and now insure around 3.5 million vehicles. We’re planning to reach five million vehicles with telematics insurance. In total we insure around 9.2 million vehicles in the Italian market. This means we would like to arrive at a bit more than 50 percent.

Behavioural tariffs usually focus on young drivers, which is very exciting but unfortunately, it’s a niche, and to cater for this group requires fairly big investments. We did turn young drivers into profitable clients, but what we would like to do is to turn telematics into a mass-market product. By working on settlements you can change the asymmetry of information.

Telematics can help avoid paying fraudulent or incorrect claims. Imagine an intersection where one driver passes a red light, the other a green one, no witnesses. Both say they had green. As a result, the insurer of each driver pays half of the cost. If you have a geo-localised vehicle you can demonstrate who passed the green light. This also benefits insurance clients as they avoid higher rates after an incident.

The other aspect we are working on is security. Imagine if you crash your car into a tree and you are unconscious. If you have a black box, our call centre phones your mobile, and if you don’t answer, we send an ambulance. This can save the client’s life.

The other aspect is more of a gimmick, but it is very popular. It allows the client to access data on their journey, for example where the nearest parking area is.

The emergence of self-driving cars poses additional challenges to the motor insurance market. And the question will be, who is liable in case of an accident? It could be that third-party motor liability becomes strictly a product liability insurance, general liability or a liability with a cyber attack or anything like that.

Selim Cavanagh: There will be a regulatory issue. The UK government is looking at how to allocate the risk.

Grange Turner: My impression is that the UK government has no intention of transferring all liability from the road, from the driver, to the manufacturer of the car in the foreseeable future. We’re not using technology to devolve responsibility entirely to a machine driving the car. You’re still going to have some human responsibility involved.

Cavanagh: They’re talking about a split policy, aren’t they?

Turner: Yes, but there is also discussion around whether the driver of the car continues to have the policy, and then there’s an easier route to subrogate against the manufacturer in the event that it’s clearly the brain of the car that’s made the error rather than the actual driver.

Freddy Macnamara: Well there are two different options here. There is a functional error if the system did something that it’s not meant to do. And then there is the possibility that the system worked perfectly well, but unfortunately, the outcome was bad. In that second situation that’s just a reinsurance problem and in the first situation, it is a warranty problem.

Dye: Under current government proposals the motor insurance policy will respond to injury or damage whether a vehicle is in an autonomous or non-autonomous mode. In the former case, the motor insurer is expected to have a right of subrogation against the vehicle manufacturer. There is clearly a huge amount of detail to be worked through in terms of what data will be available from the vehicle to establish which mode it was in at the time of the collision. This needs to be carefully worked through to ensure the subrogation process it efficient and does not add additional cost which would ultimately be paid for by customers.

Cavanagh: We’ll have many years of a mixed market, though. Adjusters may be arguing for years, but data will solve that. You have to be collecting data for these vehicles all the time, whether it’s through a third-party device or just connected though the car. You need to collect and hoover that up and try to work out who to allocate responsibility and risk to.

Mohammad Khan: An interesting question is who owns the data. Is it the car manufacturers or the consumer? If there is an accident in a driverless car currently it is only the manufacturer that has access to the data—will this continue?

Also, the potential customer base is changing, even if we ignore telematics and more autonomous driving. In the past, expensive or marque cars were a symbol of status. People wanted to own cars as they enjoyed driving. For many of our graduates now, that’s not what they’re interested in. The car, if they have one at all, is a form of transportation, a method of getting from A to B. Do they want to own it? Do they need it all year round, or even at all?

Dye: Previous estimates have suggested that for every new shared vehicle it takes between 10 to 20 vehicles off the road. That’s potentially a huge impact on the industry.

Turner:And it becomes more pronounced as we become a more urban-based society. Children growing up in central London don’t need a car: good public transport and access to Uber are cheaper than owning a car, and probably just as convenient. The idea that vehicles can be shared in that kind of communal approach to transport is absolutely prevalent.

Macnamara: The decentralisation of ownership will not benefit the zipcars in this world, the massive fleets, it’s going to be very small communities who all have access to the same vehicle and we’re trying to solve the insurance problem for it. We think they could go and sort their own car out, but the real problem is actually making sure that the insurance risk is properly protected when they’re using that car. We are building the tools to help people do that.

Dye: While you are challenging the traditional way of underwriting the industry must look to innovate.

Macnamara: One of the pieces of software we’ve built into the app recently promotes the client who insured the car through the app to make it available to their Facebook friends, who will be able to see your car on their app. Half of the problem is knowing which cars are available to use in some of these micro communities. We think that’s going to be pretty huge.

Trust is important. Over the last five or six years we’ve seen a lot of attempts to build peer-to-peer car rental communities in the same way as Airbnb. Unfortunately, that trust barrier has to be so much higher when you hand over the keys of your car. Is it going to be returned? Am I going to get a dent here? Is someone going to vomit in the back seat? But if you form trust around communities, friends, the trust is already there.

Turner: All the traditional ways that insurers have priced motor business have been on age, where you live, occupation, maybe even the colour of the car—a whole load of things which are not the risk in itself, but are pointers to what kind of risk you might present.

Presumably we’re entering a world where all those proxies become pretty irrelevant when you’ve got data on how somebody actually drives.

But who owns that data? If it’s you as the driver who owns it, you shop around with your data, you can say ‘here’s my driving profile’—it doesn’t matter what car you drive or where you live.

Dye: Independently of whether we believe that fully autonomous vehicles will dominate the future in five, 10 or 20 years’ time, the safety of cars will undoubtedly continue to develop. The driver behavioural part will become less relevant from an underwriting perspective.

Macnamara: If you take it to the extreme you’re just watching the car and if it drives through a dodgy area it pays a few more pence in premium. Because the only risks are external.

Dye: One of the interesting trends we have seen is that while the number of deaths and injuries on British roads has dropped considerably in the past 15 years the cost of injury claims is exponentially greater because of medical advancements and people are expected to live longer. This is therefore driving the need for insurers to finance their cost of care and loss of earnings for a much longer period.

Turner: You’re going to get this dynamic where there’s a frequency reduction that’s really quite marked but a severity increase which is going in the entirely wrong direction.

What is the role of the manufacturers going to be? Tesla has said ‘we’re going to insure ourselves, we don’t need insurers’.

Cavanagh: We did a study and some of the manufacturers say they will control the insurance in a one-to-one relationship with an insurer. That’s a model Tesla will have in a couple of countries around the world. But we think eventually all manufacturers will have to make that data available.

That will happen because the customer will say ‘I own the data’ and the regulator will agree and the politicians will agree and essentially most markets are competitive for insurance, with insurers and retailers competing for business. In the UK market that functions really well from a market perspective, with aggregators making a very price transparent market.

Will pay-as-you-go motor insurance become a mass-market product?

Macnamara: Our mission is to build a flexible insurance you control and the price you see on the market today is a result of that mission. We’ve been selling insurance for only a year and a half. We’ve got two products used by thousands of people. The first one offers cover for sharers. If you need to borrow a car we’ll sell you an hour of car insurance.

The second product we went live with in March allows the client to pay £20 a month so the car is insured on the side of the road. Every time the car is driven the client can buy a couple of hours of insurance or a day of insurance.

Sordoni: I don’t think it’s sow different from what’s already happening. There are many insurance products that are working that way.

With a black box in your car, if you’re just driving on the weekend, the price of your cover will be linked to the driving routine.

Macnamara: With the telematics policy, it’s very difficult. It’s not a good structure to deal with lots of people driving the same car. What we’re moving towards is you get in the car, you get your price and you can improve your price by driving better and your data is held completely separate from the car. You can get in any car and your data profile follows you because your figures are following you, and you can build up reputation based on your actual behaviour and you don’t have to proceed on a load of proxies.

Ian Brown: What’s important is knowing your customer and how that customers understands the data. If I want to get a better price by ‘driving better’, what do I have to do? What is driving better against the data and the terabytes? There’s no point in insurers saying ‘I’ve got terabytes of data which say you’re a bad driver’. What does that mean? Why am I good and why am I bad? What do I have to do?

Cavanagh: You have to focus on the big issues. We’ve got something like 130 factors in our app about driving behaviour and we probably use only about nine. The big things are distracted driving, speeding, excessive braking, that kind of stuff.

Customers are going to look at that and say, ‘it’s something I understand and I know it’s big’, and ‘it’s something I can act on and change my behaviour’.

Moger: It’s not necessarily about where you speed, it’s when you speed and in what circumstances. If you’re on an open road, it’s dry, there’s no-one else there, you do 80mph down a three-lane motorway—it’s breaking the law but it’s not dangerous unless the wheel blows. But if it’s chucking down with rain and there’s a hurricane coming from the side it’s probably not the sharpest thing to do.

Turner: The immediacy of the feedback is important; there are some telematics products where I think the insurer only gives you a report every three or six months and then if you’re not performing they won’t renew with you, or in an extreme they might cancel. That type of approach doesn’t bring a very strong feedback loop.

Cavanagh: We did the gamification with an insurer in Australia, with an app, there was a $100,000 reward for the safest driver in Australia. All their existing customers drove really safely for a while, and there were no claims, so there are definitely good results to be had. A lot of these things are one-hit wonders—you can’t keep offering $100,000—so you have to come up with sustainable long-term strategies around the feedback and reward, etc.

Won’t cars become so intelligent at some point that they are going to correct driving behaviour?

Brown: It’s got to have some impact and improvement, things have advanced in the safety of cars and people drive better. Having anti-lock brakes means people brake a lot better than they used to.

Khan: You can imagine a scenario in the near future where some manufacturers start offering cars with all the bells and whistles in terms of the safety aspects for parents to buy their children.

Sordoni: You’re talking about the technology of the mobile phone which is an issue because the precision of the gyroscope and the kind of information they are transmitting is not as efficient as we get from the black box.

Cavanagh: It’s a multi-device market at the moment where people are coming up with new devices all the time. Black boxes probably will only ever be niche in the UK market, which is for young drivers, because it’s a very robust way of collecting things, and very trustworthy, being hardwired into the vehicle.

That will change over sectors and other groups of people, and I think it’s worthwhile collecting data in other ways but there’s definitely a place for the smartphone.

Turner: Presumably the advantage of smartphone collection is that the ownership of the data is far more likely to rest with the driver than it is with the manufacturer of the car.

Cavanagh: They’ve already spent the money on the device and the data transmission, so that’s done. Everything you do then is about integrating the app and the kind of backend communication piece and the rewards and that kind of interaction with the customer. So, you’ve cleared away a whole load of costs, and we know that our market is massively price-driven.

It seems that motor insurance will be all about data. Do we need legislation to determine who owns it and how it has to be shared and with whom?

Macnamara: We have some laws coming in pretty soon that require us to share pretty much all of our customer data. The customer can say ‘give me all the data you have on me’, and we have to hand it over to them in a machine-readable format.

Moger: Also, you’re only allowed to hold data that’s relevant, unless they give you permission to hold any historic data. Some of the stuff I was reading lately says that even policy numbers count as customer data. It’s going to be incredibly confusing.

Macnamara: We’re required to hold all that data anyway, so there’s two different laws coming in that basically conflict.

Dye: Under current government proposals the motor insurance policy will respond to injury or damage claims for whatever mode, whether it’s autonomous or manual. What we’re lobbying for, and I think the Association of British Insurers is also lobbying for, is for an efficient subrogation process. When an accident occurs and it’s in autonomous mode, it’s the fault of the manufacturer, whether it’s programmed to do what it did or whether it’s a fault.

The customer and/or the injured third party gets compensated by the motor insurer and the insurer then subrogates from the manufacturer. How do we define this manufacturer responsibility, without the availability of that data, is about the transparency and the speed that that data is provided. That’s my perspective, how we need that information and ideally it needs to be legislated.

If there is not a well-defined subrogation process what’s going to happen? You may end up with a protracted subrogation, adding additional costs.

What do you think is the biggest challenge motor insurers face?

Sordoni: We need to find a common structure to share data and we need to agree to collaborate on this.

Macnamara: The biggest challenge facing the motor insurance industry is that it will be gone in the next 20 years. There will be a bunch of reinsurers and a bunch of car manufacturers and we will have mostly autonomous driving within that time period. Whoever chooses to own a car, the insurance cost will be thousands of pounds, so it won’t be possible to drive it.

Selim: For me the biggest challenge is the fragmentation and the complexity of that industry for the next 20 or 30 years. It’ll be a mixture of semi-autonomous, autonomous and totally manual vehicles producing all sort of different information. How do insurers price that, segment it and also make a profit in the next 20 years? How do insurers collaborate and share in such a way that it makes the whole industry succeed?

Moger: The biggest challenge is underwriters changing how they view the risk and being able to find a way to utilise the data, without being scared of it and what it can reveal. All technology is there to enable us to price really efficiently, as long as it’s done proactively, and to reduce risk. I think there is a lot of inertia with underwriting.

Brown: For cars and all products, distribution is going to evolve. We’ve got to keep an eye on the end customer and how they want to do things at all times.

Turner: One of the conundrums that come with safer cars is what it does for the trade-off between frequency and severity of risk, and I do think there’s a potential for peak risk associated with motor potentially increasing. Not only will individual accidents get more expensive to compensate, but I think there is an increasing risk under this new world of autonomous cars. Lots of things can go wrong at the same time—it could be a hacking exposure, a system failure exposure, whatever it might be, but then you end up with much greater catastrophic exposure.

Khan: Eventually we won’t have motor insurance in the way that we understand it now. If someone needs car insurance will they go to a “traditional” motor insurer/broker/aggregator website in the way they do now? Or will they buy it from a social website platform such as Facebook or someone like Amazon because they already have data on the consumer and so don’t have to ask nearly as many questions as insurers currently do? The key question for insurers is how to remain relevant to customers.

Moger: Underwriters will have to find a new approach to data and risk. They will have to find a new way to address the customer’s needs, which might combine motor with travel and home insurance. How do you then get all those people sewn together to work out all the rates?

Dye: As insurers we need to develop our skillset. We need to hold true in terms of managing risk portfolios and underwriting, but we need to adapt to the change in the external environment. The way we do that is our biggest challenge and developing skills for managing data and innovating will be critical.