As insurers and investors alike get excited about new growth opportunities like cyber insurance and new parametric products, it’s easy to forget about the technologies and cultural trends that will impact traditional lines of business like the old favorite: auto.
Executive SummaryFully autonomous Level 5 vehicles won’t be here tomorrow, but understanding the future of insurance in a world of autonomous vehicles is highly relevant for carrier executives today, writes Matthew Jones, a principal of Anthemis, a financial services investment and advisory firm. Here, Jones weaves descriptions of Anthemis investments in the mobility, data and on-demand insurance spaces to examine burgeoning technologies and cultural shifts in an effort to help insurers forecast where they will remain relevant in the world of autonomous vehicles. Failure to adapt could mean that risk analysis and risk management is retained entirely “in-house” within units of autonomous vehicle software companies, he writes.
According to the Insurance Information Institute, the average U.S. auto insurance expenditure grew by 17.7 percent between 2011 and 2016 compared to -0.4 percent between 2007 and 2011. Those figures point to a robust market, and it can be difficult to see how experts are predicting the demise of insurer’s auto portfolios.
Where will change come from?
Most carrier executives point to autonomous vehicles as the industry’s key prospective destabilizer.
Approximately 18,000 drivers are killed on U.S. roads annually. Over $1.6 billion was invested in autonomous vehicle-related technology in February 2019 alone. Level 5 autonomous vehicles won’t be here tomorrow, but understanding the future of insurance in a world of autonomous vehicles is clearly highly relevant for carrier executives today. Many aspects of this upcoming shift are hypothetical today, but by closely examining the nascent technologies and cultural shifts in progress, we can begin to foretell how insurance and insurers might—or might not— remain relevant in a world of autonomous vehicles.
Insurance and Cars, They Are A-Changin’
The first obvious change we are currently seeing is the move toward extreme personalization. Like asset management, Netflix recommendations and exercise plans, customers are expecting their coverage to be tailored to their individual needs and behaviors. For example, if I drive a few hundred miles per year at relatively low speeds in a quiet local area, my policy should adapt accordingly.
The use of data directly from telematics devices is already apparent. Companies like Xapix are already improving data flow within and between automotive solution providers. In the future, every autonomous vehicle will be a connected vehicle. Each vehicle will enable personalized risk assessment. The days of the simple annual auto policy may be numbered, but how many insurers are ready for such a fundamental shift?
The second ongoing change, perhaps less obvious, relates to auto risk. It too is evolving. Today’s vehicles are equipped with advanced driver-assistance systems, and this technology is already fundamentally changing the nature of the risk inherent in using a vehicle to get from point A to point B, in all manner of conditions. Previously, driver assistance meant adaptive light controls and tire pressure monitoring. Continuing advances in computer vision technology now make it possible for companies like Humanising Autonomy to detect pedestrians and understand their intent, alerting drivers—in real time—of safety risks and significantly reducing collisions.
These technologies presage autonomous vehicles. If Silicon Valley gets it right, then augmenting these driver assistance technologies with artificial intelligence will beat human performance, reducing both risk and claims. Autonomous vehicles don’t submit fraudulent claims, either. In the context of reduced risk, premiums will shrink—gradually, at first—especially as carriers seek to cut prices to remain competitive in an environment of ever-improving safety.
What Did Swiss Re and BMW Do?
In September 2018, the two companies announced that they would work together to develop a vehicle-specific insurance rating that primary insurers around the world can use to calculate car insurance premiums. Moving beyond insurance ratings based on driver’s age or a vehicle’s horsepower, they have developed an algorithm capable of representing the complex effects of driver assistance systems on the safety of BMW vehicles as a score. This ADAS risk score facilitates calculation of an individual vehicle-specific insurance premium.
The third key trend relates to the changing dynamics of distribution. Today, in seeking to buy a new auto insurance policy, I call my preferred insurance broker, visit a price comparison website or call directly. But in a world of autonomous safety, I don’t need to be responsible for my insurance, because the risk isn’t caused by me as a driver.
Mobility as a service (MaaS), a subsector of the wider mobility ecosystem, includes the emerging trend of car subscriptions and ride-hailing. BMW’s subscription service includes maintenance, roadside assistance and, yes, insurance. That premium goes to a captive insurance company owned by BMW, and that is a premium that would have otherwise flowed to a regional or national carrier.
It’s still early days for these subscription products, but they do appear to be gaining popularity as some customers no longer want the responsibility of owning a car. Lyft’s recent S-1 filing disclosed its solution for insuring drivers: Pacific Valley Insurance Co., another captive. Lyft makes it clear that it also works with outside insurers, but if ride-hailing continues to grow, or if autonomous vehicles are only available as part of a subscription (and who wouldn’t want the latest safety updates?) or through shared mobility platforms, insurance could end up being concentrated within a handful of captives or one of a few partner carriers. What will be left for the rest of the industry to pick up remains unclear.
Follow the Money—and the Risk
On top of all these ongoing changes already in progress is the emergence of autonomous vehicles themselves. Autonomous vehicles are never tipsy, drowsy or distracted by sending a text. Neither you nor I are driving them; we’re leaving the hard work to an algorithm. This begs the question of who takes on the liability, and Volvo has been very clear indeed: “Volvo will accept full liability whenever one of its cars is in autonomous mode.”
Whether Volvo retains the risk in a captive or works with one or two carefully selected partners, this represents a shift away from insuring the human driver. The human element is no longer interesting to the insurer; the customer of the future is the developer of the self-driving software. In this new world, what should carriers prioritize and what should they invest in? What are the opportunities for startups, and what are these new insurance customers looking for?
As a priority, carriers need to invest in new systems to assimilate new sources of data in real time, to assess risk as it occurs. Startups such as Flock are pioneering this in other lines of business, such as drone insurance. This risk analysis will occur at both the micro level (Where is a given vehicle, what is it doing?) as well as the macro level (Where is my fleet and what is the risk vs. another fleet?). This is largely a technological problem since most existing core systems are unable to cope with this kind of innovation, but there is also an issue of cost. Small, highly personalized policies cannot be loaded with large fixed-cost allocations. Existing (outsized) cost bases, driven by legacy platforms struggling to keep up with the pace of change, present an enormous opportunity for startups to address.
As risk management technology advances beyond today’s driver assistance systems, carriers must develop their understanding of how specific technologies are changing the risk attached to a vehicle. Swiss Re and BMW have already realized it is no longer sufficient to know only the manufacturer and model, and competing frameworks will emerge. (Editor’s Note: In September 2018, the two companies announced that they would work together to develop a vehicle-specific insurance rating that primary insurers around the world can use to calculate car insurance premiums. The ADAS risk score takes integration of safety-relevant driver assistance systems into account.) The exciting opportunity here is for startups to be able to deliver this kind of intelligence to carriers. A truly future-proofed carrier must understand, to the extent possible, a given developer’s full software stack; this will include the effectiveness and reliability of specific layers, as well as an understanding of the software developer’s own supply chain. Which best-in-class partners do they wish to solve the hardest problems? How often do they update their software, and how does performance differ across conditions and geography? This kind of valuable data remains largely untapped.
- Extreme personalization for customers.
- Cars in a pre-AV stage with driver assistance systems detecting pedestrian behavior.
- Insurance distribution.
- Mobility as a service—subscription services and captives.
- Risk assessment as it occurs, like Flock’s drone insurance.
- Insurance rates that incorporate ADAS risk scores for vehicle-specific premiums.
- Carrier partnerships with risk originators.
Finally, as outlined above, access to these risk pools is arguably going to get harder. Carriers must be able to easily build relationships and partner with the risk originators themselves. How many carriers have partnerships with OEMs in place today? In many cases, carriers are starting with a blank piece of paper, and collaborations will involve the development of new organizational structures and teams capable of translating between the two (often very different) worlds of insurance and technology. One well-known example is how Waymo, Google’s self-driving car unit, has partnered with Trōv, an InsurTech startup. (Announced in December 2017, plans called for on-demand insurance provider Trōv to offer trip insurance customized for passengers of the service, with the coverage actually underwritten by a nonadmitted affiliate of Munich Re.)
The insurance of an intangible asset like self-driving car software represents one of the biggest potential changes to auto insurance ever. With drivers most likely no longer having a connection with an auto insurer, there is a new customer in town. Startups and carriers alike have work to do in developing clearer value propositions ready for this new normal. On the startup side, this means a strong understanding of risk and state-of-the-art technology. On the carrier side, this means partnership capabilities and low costs of production.
Neither carriers nor startups alone have all the pieces of the jigsaw puzzle, so partnerships will be key. Incumbents bring licenses and the context of how auto risk has changed over the past 100 years. Startups bring innovation and the context of how auto risk is going to change in the next 10 months.
Failure to adapt could mean that risk analysis and risk management is retained entirely “in-house” within units of autonomous vehicle software companies. We have already seen that some software developers have the appetite to retain ownership of this risk, and this appetite may grow as their confidence in their technology also grows. It is possible that risk retention is designed to be an early signal of their confidence in the software, but the traditional insurance industry should sit up and take notice.
Despite this tremendous potential for change to insurers’ portfolios and business models, the insurance industry can be a valuable partner, helping to bring the prospect of autonomous vehicles to life. It would be a shame for the industry to miss out on this exciting new future. Through greater exploration and experimentation with partners, insurers can begin to position themselves today to profit from the exciting advances to mobility on the horizon.