Driverless Cars: Why We Don’t Need a New Insurance Mechanism

Driverless cars got a fair share of negative press earlier this year, when an Uber vehicle and a Tesla Model X, both equipped with some kind of autopilot mode, were each respectively involved in a fatal accident.

But compared to fatalities when human drivers are at the wheel, the danger of autonomous vehicles is slim.

Consider that 2016 alone marked 37,461 motor vehicle fatalities—a number which has been rising steadily since 2013, according to the Bureau of Transportation Statistics. And according to a study by the Insurance Institute for Highway Safety, Google’s nearly 60 driverless vehicles, which have covered more than 2 million miles, have been involved in fewer than two dozen collisions—none of which were caused by autonomous technology system failure. In fact, KPMG estimates that by 2050, autonomous vehicle technology will result in a 90% reduction in auto accident frequency.

If all goes according to plan, won’t the U.S. auto insurance markets as we know them shift dramatically away from individual liability policies and toward product liability solutions in which manufacturers bear the brunt of the exposure for at-fault accidents?

Not according to the Travelers Institute, the public policy division of the Travelers Companies, Inc., which recently released a white paper assessing existing personal and commercial auto insurance structures and how they could be applied to a future dominated by autonomous vehicle technology.

IA spoke with Michael Klein, executive vice president and president of personal insurance at Travelers, about the timeline for autonomy, the implications for both the personal and commercial auto insurance markets, and how insurers and agents will need to respond and adapt.

IA: Considering the fact that the average U.S. car is 12 years old and that only 5% of the current fleet incorporates driver assist technologies, what’s a realistic timeline when we’re talking about autonomous vehicles?

Klein: There are a number of assumptions underneath some of those very aggressive projections as to how rapidly private passenger auto premiums will diminish. One is the assumption about reduced frequency and improved safety as autonomy progresses, and we believe in that. But there’s also a whole set of assumptions around how rapidly the technology progresses, around whether the infrastructure keeps pace, around the rate of adoption, and one key thing to think about there is affordability. Assumptions that tend to be consistent with the more rapid acceleration assume that there aren’t going to be barriers to adoption, but there are plenty.

The last assumption is that there’s going to be a significant decrease in vehicle ownership, either because people are ridesharing more frequently or they’re deciding they don’t need to own as many cars because the autonomous one will be able to serve multiple people in the household at once. And that could potentially be the case down the road, but many of these predictions tend to lean on all of those assumptions lining up, and that’s how they get to that end of the spectrum. That’s also why there’s such a wide degree of variation in the estimates we see out there.

Based on the statistics you cited, even if every new car shifts to fully autonomous in 2021—which is the year most manufacturers are talking about targeting—it would take 12 more years for just half of the fleet to be fully autonomous. And neither of those things is true—they’re not all going to be fully autonomous, and not every single car will be shifting in 2021. So I think the timeline for any sort of a tipping point is a long way off.

All that said, you only have to look at some of the recent headlines to see that we’re already in a world where in the case of an accident, there is some question as to whether it was the driver’s fault or the vehicle’s fault. So even though the tipping point is a long way off, the need to deal with the insurance implications of autonomous vehicles is already here. That’s one of the things that drove us to publish the white paper in the first place—one of the potential barriers to that progress is this perceived need to invent a new approach to dealing with the insurance implications, and our perspective is we don’t need to invent a new approach.

Walk me through that—that’s a pretty unique position compared to the rest of the industry.

The fact of the matter is if somebody buys a car with autopilot mode today, they have an auto insurance policy that covers physical damage and liability. If they’re in an accident and it is the vehicle’s fault, the insurance policy will respond and get their car fixed and get their injuries taken care of, and then the insurance company will subrogate against the manufacturer. There’s already a process in place in the market to deal with the increased autonomy, and we think existing products are already well-positioned to respond to those dynamics.

Product liability often involves complex, lengthy lawsuits, which is why we do not believe product liability is the right mechanism for responding to auto accidents. Auto insurance is an efficient mechanism. Auto insurers have claims adjusters deployed throughout the country, and they continue to invest in capabilities to more rapidly and efficiently adjust auto claims. If we use the existing mechanism to apportion liability behind the scenes, we can take care of drivers and vehicles more efficiently and more effectively—get the cars fixed, get the injuries taken care of, and help people get on with their lives and their cars back on the road.

Even if we don’t need a new mechanism, how does auto insurance need to change to account for not only more expensive, sophisticated technology in autonomous vehicles, but also the various layers of liability that will result from accidents when the driver does not have control over the vehicle?

That’s a good distinction—the mechanism will certainly need to evolve. It will have to be the responsibility of the owner to have sufficient liability limits at the vehicle level to respond to accidents. Today, we often think about liability going with the driver, but as autonomy increases and there becomes more question about whether it was the driver’s fault or the car’s fault, placing the liability on the owner of the vehicle makes a little bit more sense.

And as vehicles grow in their autonomous capability and continue to get more expensive to build and repair, we think it’s even more important that there be enough liability limit there to cover that property damage exposure that’s getting bigger.

Can you discuss some of the differences between autonomous technology development in the personal auto space versus the commercial auto space?

In some ways, we expect progress on the commercial side to be quicker. When you look at some of the developments in autonomy, those are easier problems to solve on the commercial side, because you can build and design a vehicle to deal with whatever variations it’s going to encounter in more of a fixed-route operation. That probably won’t include an individual’s driveway and garage, for example—getting in and out of the garage and out onto the street is going to be very unique from car to car.

Or think about over-the-road trucks. There’s already been quite a bit of progress in autonomous trucks for limited applications. A lot of that testing has been on interstate highways that frankly are some of the best infrastructure we have in the U.S., and probably are some of the most logical and safest places to do testing and apply the technology—those straighter interstates in the middle part of the country, for example, as opposed to some of the highly congested areas in the Northeast. You’ve got thousands of miles of interstate in the which, in terms of an autonomous vehicle being able to navigate it, is a more straightforward application.

What kinds of regulations will need to be implemented so that driverless cars can be integrated into current state-based traffic laws and oversight structures?

First and foremost, when and where can autonomous vehicles be tested, and what are the requirements around testing? Then, there’s a whole set of responsibilities around data rights and data exchange, because if you’re going to be able to do a full diagnostic on who was at fault, the data in the vehicle is going to be an important piece of that process. You’ve also obviously got insurance liability-related regulation, and you’re going to have to have regulations around the capabilities of the equipment—manufacturing standards and requirements around safety.

Finally, you’re also potentially going to have to change the requirements around road design to ensure that the infrastructure necessary to support autonomous vehicles is there. Autonomous vehicle technology is progressing rapidly, so I’m sure it will eventually evolve to be able to better respond to the infrastructure that exists today. But to the degree that the infrastructure can be improved and adapted to support increased levels of autonomy, I think the two can go hand in hand.

At what level do you expect regulations to originate? Local? State? Federal?

I think it’s going to be quite a mixture. It’s Travelers’ position that a state model law regarding the insurance issues would be helpful. In many cases, regulations around testing and safety requirements often come from the federal level, but there’s a whole set of individual state and sometimes even local regulations and permissions and approvals that are happening today. Our focus right now is predominantly on the insurance-related issues, and we believe that’s most appropriately dealt with at the state level.

Based on your research, do you believe the independent agent has a role to play in the personal auto market of tomorrow?

For the foreseeable future, there’s still an important place for agents in this space, if for no other reason than in times of uncertainty, people look for advice, and agents are in a position to offer that advice. Now, I think it’s really important that agents stay educated and informed on these emerging issues. As you think about a potential for either driver assistance features or autonomy in vehicles to change the nature of exposure and change the risk agents’ customers are facing, that’s actually precisely the time agents need to be there to advise them on the right decisions. There’s lots of room here for all of us, carriers and agents alike, to work together to try to help drive toward the best solution.

Jacquelyn Connelly is IA senior editor.