Content

Future of Personal Lines Auto Insurance: When and How Might the Industry Transform Itself?

Dr. Anand S. Rao

“Driving a Car will be Illegal by 2030,” declared Wired earlier this year¹; “Are driverless cars an auto insurance train wreck” asks Bloomberg²; “How the Autonomous Car will Change the World and Upend Auto Insurance”¹⁷ explains Brad Templeton from the Singularity Institute.

This article takes a look at some of the major trends in the sharing economy, automated driver assistance systems, connected cars, and autonomous cars, and assesses their potential impact on road safety, the frequency and severity of accidents, customer adoption of new technologies, and regulatory changes. We also describe an advanced model that facilitates projections of different alternative future scenarios to estimate risk in an auto insurance market with semi-autonomous and autonomous vehicles. Irrespective of the speed at which some of these changes will manifest themselves, insurers can prepare for them by forming alliances with auto manufacturers and others outside of the insurance industry to collect the data and make informed choices on how to appropriately tailor their products, pricing, underwriting, claims, marketing, and distribution.

Forces of change

While autonomous vehicles (AV) or driverless cars get the lion’s share of media attention, there are at least four additional technological developments that are reshaping the auto sector: connected cars (i.e., Internet enabled vehicles), automated driver assistance systems (ADAS), car/ride sharing, and alternative fuel vehicles.

Connected cars can further digitize the auto sector and foster ADAS advances: Connected cars currently tend to feature only navigational and infotainment services. However, diagnostic information from vehicle performance can be useful for a variety of purposes, including predictive maintenance of different parts, understanding and modification of driving behavior, claims and liability assignment, and warranty issues. As technology and customer adoption improve, connected cars and the broader digitization of driving are likely to have a significant impact on the auto and auto insurance sectors.
ADAS can influence customer acceptance of new technology and the path towards autonomous vehicles: The US National Highway Traffic Safety Administration (NHTSA) recently came up with five levels of maturity for automated driver assistance³. The five levels provide a useful framework for examining the different types of technologies—Level 0: No automation; Level 1: Function-specific automation e.g., cruise control, automatic braking, and lane keeping; Level 2: Combined function automation e.g., adaptive cruise control and lane centering; Level 3: Limited self-driving automation where the driver cedes control of all safety critical functions under certain traffic or environmental conditions; Level 4: Full self-driving automation where the car performs all safety critical functions under all conditions.

In order to assist drivers, various auto manufacturers are deploying a number of in-car technologies, such as forward collision warning, drowsy driver detection, adaptive headlights, lane departure sensing, blind spot assistance, parking assistance and adaptive cruise control. When motorists use some of these automated driver assistance technologies in tandem, self-driving automation can reach Level 2 or even Level 3. These ADAS features are becoming more widespread. As the customer adoption and penetration of these technologies increase over the next five to 10 years, customers will be more open to “handing over control” to automated systems, paving the way for autonomous vehicles.

The sharing economy has the potential to drive AV adoption: Increased urbanization and congestion, and the high total cost of vehicle ownership is leading many urban dwellers and Millennials to increasingly adopt car sharing/ride sharing. The popularity, ease of use, and convenience of these car sharing/ride sharing services is increasing vehicular leasing and rental on a “per trip” basis. This alternative “personal mobility” market is creating an opportunity for service providers to innovate and introduce ADAS and eventually autonomous vehicles into the market in order to maximize asset utilization and decrease cost of maintenance/ownership. Autonomous vehicles will be able to service more daily consumer trips within a sharing economy (i.e., every autonomous vehicle will displace nine traditional vehicles in a sharing fleet²⁰). This will allow service providers to increase service convenience and attractiveness at equivalent costs.
Electric and alternative fuel vehicle adoption could increase with more AVs and ride sharing: Electric vehicle adoption could benefit from the synergy between any decline in car ownership and increased adoption of car sharing. Autonomous technology have the potential to introduce efficiencies that enhance electric vehicles’ attractiveness and affordability resulting from lower cost electric component configuration, increased range, and faster payback periods. In short, technological advances are likely to influence customer adoption and in turn be influenced by it. Not all technologies will find widespread customer adoption, but those that reach a certain critical mass will reduce technology costs and increase affordability and acceptance.

Regulatory change is likely to lead to greater manufacturer and customer adoption of technological change: As of late 2015, 16 states had introduced legislation related to autonomous vehicles. In addition, the US Transportation Secretary has committed $4 billion over the next 10 years to accelerate the development and adoption of safe vehicle automation through the National Highway Traffic Safety Administration (NHTSA). Legislative steps coupled with funding for safety related automation likely will mandate at least some ADAS technologies in new vehicles. In turn, this likely would lower costs and further technological advances.

Potential impacts on the auto industry and driving behavior

The complex interplay of technological advances, customer adoption, and regulatory changes in the auto sector is difficult to estimate or project using traditional techniques. We have developed AutoSim, an advanced agent-based simulation model that uses advances in artificial intelligence, to develop a deep causal model of all of the interacting factors. The model’s objective is to evaluate a range of alternative scenarios with outcome variables that include rate of technology advances, rate of customer adoption of new technologies, asset utilization, impact on frequency and severity of accidents, and pace of regulatory changes. Key modelled factors include:

Technology Impact—Different auto manufacturers implement technologies such as forward collision warning, drowsy driver detection, and adaptive headlights in different ways. As a result, reductions in collision damage could vary depending on the effectiveness of respective implementations.
Availability—Depending on manufacturer cost technology, regulatory requirements, and customer adoption, automated technologies may be available as standard or optional features. In addition, auto manufacturers may deploy these technologies on just a few higher end models or on all of their models.
Consumer Behavior—These are the choices that individual consumers and households make about personal mobility (e.g., buying, leasing, renting a car, ride sharing, walking, or using public transport). These choices impact the extent of car sharing/ride sharing adoption, as well as adoption of new ADAS features and AVs.
Usage—A human driver has to activate some of the technologies, like adaptive cruise control and parking assistance. As a result, even if vehicles feature them, drivers may not use some or even all of the technology at their disposal.
Regulatory Intervention—Given these technologies’ potential benefits to safety, congestion, and greenhouse gas emissions, the government is likely to mandate their eventual usage. However, it is also likely that it will want to rigorously test them under all conditions before they approve widespread deployment. This could either speed up or slow down adoption of automated technologies.
Penetration—The availability and use of automated technologies will change over time as they mature and drivers become more comfortable using them. The percentage of new cars with these technologies will be a critical factor in their overall impact.

We used AutoSim to model the frequency and severity of accidents using a three-step process. The first step was Technology Impact Analysis, where we used the Highway Loss Data Institute’s detailed research¹¹ to estimate the impact of 11 different automated driver assistance technologies (e.g., forward collision warning, drowsy driver detection and warning, Adaptive headlight, Lane departure, etc.) on the frequency and severity of five types of claims: bodily injury liability, collision, personal injury protection, comprehensive claims and property damage liability. In the second step, Adoption Projection, we projected the impact of specific technologies on the four types of claims that we estimate will be adopted by manufacturers and drivers over the next 20 years. In the last and final step, Loss Reduction Estimation, we determined the net reduction of losses across the five major categories of claims, and use the assumptions of the adoption of automated technologies and how long it will take to replace older vehicles.

The model is sensitive to the assumptions we made along the three steps. In particular, the key assumptions that drive the overall results are:

Technology Impact—Based on our analysis of all the technologies we describe above, the reduction in losses include bodily injury (-15 percent), collision (-6 percent), comprehensive (0 percent), property damage and protection (-14 percent), and personal injury protection (-10 percent). This is an average we base on different technologies that different auto manufacturers deploy^{4,
5, 6, 7, 8, 9}. As technologies improve and manufacturers learn from their own and other manufacturers’ experiences, these impacts are very likely to change.
Availability and Adoption—Historical analysis shows a 15-year span between initial introduction of a new technology and 95 percent new vehicle availability. In addition, we assume that it takes an additional 15 years (or 30 years total) to reach 95 percent of all vehicle availability. If some of these technologies demonstrably prove to contribute to passenger safety^{10, 11, 12}, then regulatory mandate could accelerate their widespread availability. In addition, adoption is likely to come in multiple waves as different technologies are piloted, tested, deployed as optional, and finally available as standard equipment¹⁶. Frost & Sullivan estimates around 3.2 million semi-automated, highly-automated, and fully-automated new vehicles in North America and around 3 million in Europe as part of the third wave of shipments¹⁶
Baseline—We assume a linear projection of losses for the baseline based on projecting vehicle miles driven and loss experience from 2009–2013. This projection suggests that total losses would grow to $83 billion by 2025 and $101 billion by 2035 if new technologies have no impact.

Based on the overall assumptions and the analysis, we estimate a reduction of losses of around 10 percent for the U.S. auto market by 2025 and 35 percent by 2035. Combining ADAS with AV introduction, these numbers could rise to 30 percent by 2025 and 50 percent by 2035 with accelerated adoption. We estimate the net baseline projected losses without driver assistance technologies to be $83 billion by 2025, and $76 billion with driver assistance technologies. By 2035 we project losses without driver assistance at $101 billion, and $80 billion with driver assist technologies. We believe that these estimates are conservative, and if we relax some of our assumptions of the rate of availability, pace of adoption, and impact of technology on losses, then we estimate even greater loss reductions. For example, Thatcham research¹⁹estimates that 80 percent of all crashes in the U.K. occur at a speed of less than 25 Km/hr. Automated Driver Assistance Systems focused on safety systems at lower speeds (e.g., forward collision with automatic breaking, emergency brake assistance) can result in a significant reduction in accidents and injuries, and when accidents do occur, they are likely to reduce the severity of injuries.

Potential impacts on auto insurers

Although current technology and legislation are still very much works in progress, there are strong forces that do stand to reshape the sector, including shifts to new types of coverage, alternative distribution channels, and redefined customer segments. We envision five highly possible evolutionary changes to the personal auto insurance industry's products, distribution, and customers, as well as one more truly transformative change that would significantly affect the shape and size of the industry as we know it.

Risk Shifting—Automated driver assistance technologies, such as forward collision warning and drowsy driver warning, will increasingly shift the risk of driver error to the risk of mechanical malfunction¹⁵. This would shift liability to manufacturers and result in a new form of auto insurance that could be packaged with cars that rely on these technologies. In turn, this would shift the key buyer from the end-consumer to the manufacturer and fundamentally change the entire value chain, from product definition to pricing, marketing, distribution, underwriting, service, and claims. If carriers decide to market this coverage to consumers, then they would do so either at the point of sale, or perhaps try to increase market share by co-marketing with the manufacturer and/or dealer. Risk shifting offers opportunities for auto insurers to gain market share by arranging deals with auto manufacturers, but also a threat to companies that fail to capitalize on these opportunities.
Risk Sharing—Smartphone apps have already started playing a role in collision reduction. In addition, the dramatic rise in social networking has enabled individuals to develop new affinities wherein people with similar attitudes, interests, and behaviors can pool resources to share risk and lower overall costs. For example, there are new carriers that combine social networking with insurance by connecting customers to form insurance networks that promise significantly lower premiums. These carriers claim that their models allow insurers to access new customers virally, decrease process costs, and reduce claim ratios. While this represents the potential for lower rates for more people, it also could make insurance more affordable for many and therefore lead to premium growth.
Risk Slicing: Self-driving Mode—In the next five to 10 years we are likely to see more cars with a self-driving mode. Drivers will be shifting between hands-on and hands-off-driving depending on conditions. This will result in different risk profiles for a single trip and also different liabilities—driver liability in the hands-on mode and product liability in the self-driving or hands-off mode. This type of risk slicing offers a number of pricing options for auto insurers. Similar to usage-based or mileage-based insurance that telematics-driven auto insurers offer, we could see insurance premiums priced differently based on the mode of driving.
Risk Slicing: Car Sharing—An alternative type of risk slicing occurs with the growing trend of car sharing. Urban living and the increasing availability of automotive time-sharing suggests a future in which premiums move from 24-hour asset coverage to a pay-per-use model. Over 80 percent of the U.S. and over 50 percent of the global population is considered urban; understandably, car sharing is rapidly growing¹⁷. According to a Frost & Sullivan research estimate that Forbes¹³ reported in March 2012, the global car sharing market could exceed $10 billion by 2020, and the North American car sharing market alone could surpass 4.4 million members and $3 billion by 2016. In Europe, the number of members will rise to 15 million by 2020¹⁴. As a result, an increasing number of low-frequency drivers is likely to mean at least some reduction in individual premiums. However, this scenario does not necessarily represent only lost premiums. Most of the people who do not choose to own cars will need to rent them at least occasionally; accordingly, car sharing can expand the market for alternative buyers of insurance.
Risk Reduction and Elimination—Unlike the above scenarios that represent significant change but not necessarily extreme disruption to the insurance industry, driverless cars or autonomous cars (Level 4) equipped with the latest awareness technologies could completely change the industry as we know it. Google, Inc.'s auto research investments are hastening the public availability of driverless cars. In addition, Google and other companies are investing in the research and development that initially sets and then drives down the costs of new technologies¹⁷. In the U.S., driverless cars are now legal in California, Nevada, Michigan and Florida. Google estimates that the technology can reduce traffic accidents by 90 percent, reduce the number of cars by 90 percent, and reduce wasted commute time and energy by 90 percent resulting in savings of $2 trillion per year to the U.S. economy¹⁸.

Getting ready

The widespread adoption of car sharing/ride sharing, and ADAS is not a matter of if but when. On the road to fully autonomous cars are a number of risks and opportunities for auto insurers. Ignoring them or not taking decisive actions could prove fatal. Some of the ways to turn ADAS adoption into an opportunity include:

Product Innovation—Usage-based, driving mode-based, and trip-based insurance using telematics devices and ADAS offers insurers new product innovation opportunities. Insurers who are able to unbundle auto insurance and re-bundle it in new ways to target emerging urban and casual drivers, as well as self-driving car drivers.
Distribution Innovation—Rise of affinity groups, car sharing groups, and vehicle manufacturers who want to package auto insurance with autonomous vehicles can offer new distribution channels to auto insurers. Disruptive players who focus primarily on these segments can adopt a B2B distribution channel directly with auto manufacturers or their dealers or personal mobility service providers. These players would have a fundamentally different business model and could progressively capture market share as the proportion of autonomous cars increase.
Service Innovation—As the need for protection decreases, insurers can play the central role of aggregating information and entertainment needs. Auto manufacturers, online or mobile service providers, telecommunication providers, and information providers are all vying for leadership in in-car infotainment services. Insurers with trusted brands can re-orient themselves as service providers.
Claims Innovation—The biggest impact of ADAS and autonomous cars will be on safety and the prevention or reduction of accidents. Insurers who approach insureds with these types of cars as a separate segment and handle claims based on on-board diagnostics and analytics will use fundamentally different economics for claims handling and the legal expenses associated with claims. Such auto claims settlement will increase claims satisfaction and reduce litigation costs.

Given the widespread and long-term impact of these technologies to auto insurance, we recommend a three-phase roadmap for auto insurers:

Short-term (3–7 years): (a) Embrace alternative forms of insurance (e.g., MetroMile); (b) Partner with auto manufacturers to collect data on ADAS and AV technologies; (c) Build capabilities on usage-based underwriting and risk slicing; and (d) Tighten underwriting guidelines for individual consumers.
Medium-term (8–15 years): (a) Shift primary focus towards auto-manufacturers and personal mobility providers for marketing, distribution and product development; (b) Consolidate underwriting capacity as premiums start shrinking; (c) Develop innovative products—bundled insurance, “self-drive” mode insurance.
Long-term (15+ years): (a) Consolidate personal and commercial auto; (b) Explore alternative revenue streams including, commercial auto logistics, becoming a personal mobility service provider, etc.

Whatever the future holds, the automotive insurance business is going to change. Despite some doomsday predictions for the industry, there are opportunities for insurers to develop innovative new products, alternative distribution approaches, and new customer segments that can help them thrive, not just survive. The carriers that can think creatively about new markets and potentially drastic changes to automotive technology and ownership will be the ones who are most likely to successfully navigate the path to the future.

Acknowledgements: The author would like to thank Mark Paich, Lyle Wallis, Joe Voyles, Ilana Golbin, Scott Fullman, Spencer Allee, and Balaji Jayakumar for their contributions.

References

“Driving a Car will be Illegal by 2030,” Wired UK, January 2016
“Are Driverless Cars an Auto Insurance Car Wreck,” Bloomberg, Jan. 16, 2016.
Preliminary statement of policy concerning automated vehicles. National Highway Traffic Safety Administration, 2013
“Mercedes-Benz collision avoidance features: initial results.” Highway Loss Data Institute Bulletin. Vol. 29, No. 7: April 2012.
“Honda Accord collision avoidance features: initial results.” Highway Loss Data Institute Bulletin. Vol. 31, No. 2: April 2014
“Volvo collision avoidance features: initial results.” Highway Loss Data Institute Bulletin. Vol. 29, No. 5: April 2012.
“Acura collision avoidance features: initial results.” Highway Loss Data Institute Bulletin. Vol. 28, No. 21: December 2011.
“Buick collision avoidance features: initial results.” Highway Loss Data Institute Bulletin. Vol. 28, No. 22: December 2011.
“Mazda collision avoidance features: initial results.” Highway Loss Data Institute Bulletin. Vol. 28, No. 13: December 2011.
“Predicted availability of safety features on registered vehicles – an update.” Highway Loss Data Institute Bulletin. Vol. 31, No. 15: September 2014.
“Estimated time of arrival: New safety features take 3 decades to spread through vehicle fleet.” Insurance Institute for Highway Safety Loss Data Institute Bulletin. Vol. 47, No. 1: January 2012.

“V2V: Cars Communicating to Prevent Crashes, Deaths, Injuries.” DOT: February 2014.
“Zipcar fuelled up for $22 Run as Business Model Matures.” Forbes. March 20, 2012.
“Growing Awareness of Peer-to-Peer Carsharing will Boost Carsharing Rentals in Less Populated Areas in Europe.” Frost & Sullivan. Aug. 22, 2012.
“The Reshaping of Auto Insurance.” Marik Brockman and Anand Rao. Insurance and Technology. Jan. 8, 2013.
“From Vehicle Automation to Autonomous Driving: The Big Leap.” Prana Natarajan, Frost & Sullivan. Proceedings of the The Autonomous Car: Risks and Opportunities for the Re/Insurance Industry. September 2014.
“How the Autonomous Car will Change the World and Upend Auto Insurance.” Brad Templeton, Singularity University. Proceedings of the The Autonomous Car: Risks and Opportunities for the Re/Insurance Industry. September 2014.
“Fasten your Seatbelts: Google’s Driverless Car is Worth Trillions (Part 1).” Chunka Mui. Jan. 22, 2013.
“Understanding Technological Advances in Vehicle Safety to Reduce Claims Costs in the Future.” Matthew Avery. Thatcham Research. 2014.
“Disruptive Mobility: AV Deployment Risks and Possibilities,” July 20, 2015, Barclays Report.

Anand Rao is a principal in PwC’s Advisory practice, is the Innovation Lead for the US firm’s Analytics Group and is the co-lead for the Global Project Blue, Future of Insurance research. With his Ph.D. and research career in Artificial Intelligence and his subsequent experience in management consulting he brings business domain knowledge, statistical, and computational analytics to generate unique insights into the practice of data science.

Prior to joining management consulting, Anand was the Chief Research Scientist at the Australian Artificial Intelligence Institute. He has held board positions at start-ups and currently serves as a board member for not-for-profit industry associations. He is a frequent speaker on behavioral economics, analytics, and technology topics in academic and trade forums.