A whole new (autonomous) world
In March, the Bank of England estimated that the impact of autonomous vehicles will lead to a 21% contraction of the UK motor industry by 2040, so to what extent should insurers in the motor industry be concerned?
Many of the automated functions in cars today were designed to increase safety which may make consumers feel more at ease, but the verdict is yet to come in for insurers on their impact. In the last fifteen years, as automated technologies have begun to sneak into more vehicles (in many cases now as standard), the number of people killed or seriously injured in road traffic accidents has halved in the UK. With 90% of road accidents resulting from human error, if the scope for such error is reduced or removed altogether then, logically, the number of claims will also reduce. This, coupled with a reduction in the severity of claims, may push down premiums and have a negative knock on effect for motor insurance profits.
Conversely, damage to automated vehicles is likely to be much more costly to repair and parts (such as sensors) will be more expensive to replace which could push up premiums. Further, as the AA reported, a Transport Research Laboratory study found that as autonomous vehicles become more commonplace, motorists' driving habits may change detrimentally, such that drivers may take more risks and, in the short term, produce worse accident statistics resulting in increased insurance premiums for those driving ordinary cars.
Overall, it appears that insurers are supporting the move towards an autonomous automotive industry and with good reason. The government has sent a clear message: autonomous transport is the way forward and the intention is for the UK to be in the poll position. The industry has responded: manufacturers and service providers from Ford to Google to Uber, are investing in autonomous testing and development. From a driverless bus route in Paris to a self-driving shuttle bus in Germany, autonomous technology is inescapable. By the 2030s, the UK driverless car market is predicted to be a multi-billion pound industry and it is estimated that reducing road accidents through the use of automated technology, will have contributed £2 billion of savings to the economy. Furthermore, the world of InsurTech has opened up the market to innovative start-ups which are producing unique products to meet the demands of the rapidly changing landscape. Insurance firms which continue to rely on outdated IT platforms will struggle to keep pace with hungry new competitors offering new ways to underwrite, price and sell insurance. Insurers have, therefore, understandably been considering how best to position themselves as the industry transforms.
There appears to be a general consensus that fully-automated cars (i.e. those without steering wheels or pedals, nor the need for operator intervention) will not appear on the market before the early to mid-2020s, but in many ways, the future has already arrived in the shape of a whole host of autonomous features in vehicles, including remote control parking, lane departure warning systems, automated emergency braking (AEB), anti-lock braking (ABS), adaptive cruise control, blind spot monitoring, and other crash-avoidance technology.
The current pace is set to continue. A plethora of manufacturers and tech companies are jumping on board: autonomous technology is no longer the sole domain of early adopters – now companies must innovate or die. This is also true for those looking to provide products to insure against the risk: the insurance market will continue to become more competitive, not least because it is predicted that eventually all vehicles will be autonomous. A select number of insurers have already adjusted their policy prices based on the presence of autonomous or semi-autonomous features, offering discounts for certain cars' features. Root, for example, an insurance start-up, offers cheaper rates for Tesla drivers with auto-pilot functionality.
Evolution of motor liability
Driver, manufacturer or another third party?
As cars are evolving, liability is becoming an increasingly grey area. The apportionment of liability will become murkier still in instances where it is not clear who, or what, is at fault. Traditional motor insurance generally assumes that drivers will be personally liable and, at the moment, the law's emphasis is on the actions of the driver. However, as a number of commentators have highlighted, risk will start to shift away from individuals to product and software – something that legislative changes now highlight. As the Government Actuary's Department (GAD) has noted, there has been debate in actuarial circles about the complications around establishing responsibility. Actuaries will be faced with a number of new challenges, not least the need to begin weighing up entirely new factors, for example, how often cars are hacked and the reliability of certain software. The GAD has suggested that we are likely to see a transformation from personal liability to product liability.
Understandably, many manufacturers will resist this change and there is certainly an argument that where a driver is fully licensed and sees that an accident may take place, they should intervene. Of course, truly driverless cars will have neither a steering wheel nor pedals and it may be impossible for a driver to do anything at all in an emergency. In those circumstances, consumers are likely to rely on the Consumer Protection Act 1987, which would impose strict liability on a manufacturer if there were a defect in the car.
That said, other factors may influence whether a person should still be liable in a fully autonomous car, for example, a failure to service, maintain or upgrade the vehicle or its software in line with manufacturer instructions. Manufacturers will need to deliver clear communications to consumers, not only about precisely how to use and maintain vehicles, but also to manage expectations about the vehicle's capabilities.
A further argument against manufacturers accepting all liability is that it may not have been the manufacturer who created a particular element of the software which has malfunctioned. In such cases, there is an obvious argument that the software provider (and/or other relevant third party suppliers) should accept liability. Claims may well be pursued against a number of parties, which could result in drawn-out and costly litigation.
Alternatively, we may see a greater number of 'no fault' or strict liability insurance policies, whereby insurers would pay out even if an 'at fault' party is not identified. Some larger manufacturers (for example Volvo) have, however, been willing to back their autonomous systems with complete confidence by suggesting that they would be willing to accept liability.
Level of autonomy
How liability is apportioned will also depend on the level of autonomy that a vehicle provides. This ranges from partially autonomous cars (where drivers do not control acceleration or steering, but must remain alert in case of an emergency), highly autonomous cars (which are essentially prototypes and will require the driver's control at different points throughout a journey) and fully autonomous cars (which can reach a destination without any driver even being present in the vehicle). The extent to which a user can be said to be liable will differ with each.
Calculating premiums in relation to automated vehicles is a challenge, since present actuarial data such as the age and driving records of policy holders will not be as relevant given that the closer a vehicle is to being fully autonomous, the less likely the driver will be found to be at fault. Instead, the make and model of a car may become more important. As Munich Re has highlighted, underwriters will also need a detailed understanding of a vehicle's operating limits in certain conditions, for example, in poor weather.
Patrick Lin, a contributor for Forbes, has hypothesised that, given the real-time data that cars of the future could generate, perhaps real-time insurance policies may start to develop such that if a driverless car user has been drinking, or if they are driving late at night, their car may ask them to accept that, for that journey, since there is a small chance the user may have to step in in an emergency, their insurance rate will increase. Various issues would attach to this approach (those of ethics and privacy to name just two).
Achieving clarity and certainty for both insurers and consumers around criminal and civil liability in the event of motor accidents involving each of partially, highly and fully autonomous cars, is going to be critically important.
Following a Department of Transport and Centre for Connection and Autonomous Vehicles consultation, various measures informed by responses to the consultation from the industry (as regards to providing an insurance framework to cover liability for autonomous vehicles) were proposed by the government in the Vehicle Technology and Aviation Bill (previously the Modern Transport Bill). This Bill did not get through Parliament due to the early dissolution as a result of the snap election earlier this year. However, the proposed measures to extend compulsory insurance to cover the use of autonomous vehicles in the original Bill will now be taken forwards in the Automated and Electric Vehicles Bill (AEV Bill) which was announced in the Queen's Speech on 21 June 2017. The AEV Bill envisages a "single insurer" model policy and its measures will include:
- amendments to the Road Traffic Act 1988 so that compulsory insurance is extended to autonomous vehicle technology (the Secretary of State will maintain a list of those vehicles which qualify as "autonomous");
- compulsory insurance to cover both drivers and their use of the vehicle, as well as 'drivers', who are validly removed from the actual driving (i.e. when an autonomous vehicle is operating in autonomous mode); and
- first instance liability will fall on insurers so that victims of collisions resulting from autonomous vehicles will be able to get compensation easily, expeditiously and fairly.
Coverage will include both injured parties, but also property damage. There will also be first party cover for personal injury and fatalities. As well as the conditions where the insurer would be liable for damage caused by an accident, the Bill also includes limitations of insurers' liability with exclusions for where the driver was negligent in allowing the vehicle to drive in autonomous mode in circumstances where it was inappropriate, or where there was a failure to install/ update required software, or where the operating system had been altered or tampered with.
The government dropped its proposal to include an element of product liability coverage in the legislation following feedback from the industry (for example, the Association of British Insurers (ABI)) and the crash test firm Thatcham Research submitted a joint response to the consultation in which they argued against drivers having a separate product liability policy on the basis that, among other things, it would be too complicated. Elsewhere, a number of responses revealed concerns about the co-existence of product liability insurance law and the current compulsory motor insurance framework, for example, because product liability insurance policy terms are not controlled in the same way as road traffic policies, and because product liability insurance is optional. The government came to the view that amending product liability law to accommodate the developments in autonomous vehicles would (at this stage) be disproportionate (see our article for more on product liability and CAVs).
Clearly, there will be a transition period during which time the extent to which autonomous vehicles are pervasive will gradually increase. It is intended, therefore, that the regulatory and legislative reform will be introduced on a rolling basis. During such transition period, it is also expected that insurers and manufacturers will work together to share data.
Insurers will undoubtedly welcome the fact that under the AEV Bill they will be able to recover from any responsible parties (through existing common law/ product liability laws), whether such parties are manufacturers, vehicle servicing providers, or even hackers. They will also be able to pursue a product liability or professional indemnity insurer. Unsurprisingly, however, this process may be complex as a number of parties may be involved. It would also still require the insurer to prove that the third party had been negligent. They will also be relieved that the Bill provides scope to limit or exclude liability where insureds fail to install critical software updates where they ought reasonably to have known they were required or where they make unauthorised software alterations.
The UK Government continues to show its commitment to actively developing and investing in driverless technology and has been clear that it is dedicated to being at the forefront of the industry. It described itself last year as being "determined that Britain leads the way globally in embracing the safe development of driverless technology".
In addition to its commitment to address the legislative framework through the AEV Bill, it has (by way of example) also:
- published a Department of Transport Code of Practice for testing driverless cars;
- established the Centre for Connected and Autonomous Vehicles (CCAV) to ensure the UK remains a world leader in developing and testing connected and autonomous vehicles (including a programme of research and development worth up to £200 million);
- funded various driverless car projects in the UK and invested £50 million in feasibility studies;
- announced that fleets (or 'platoons') of driverless lorries will be trialled on UK motorways, investing £8.1 million into the project;
- backed a consortium (including Direct Line, FiveAI, Oxford University, Transport for London and the Transport Research Laboratory) in April this year with £12.8m to research, develop and trial driverless car systems in London – their plan is to demonstrate a fully autonomous system by Q3 2019;
- taken advice from AXA on how insurance claims may work in the future – this has involved, for example, AXA deliberately crashing cars to recreate autonomous situations to help evaluate insurance premiums; and
- awarded £51m to four projects across five locations, to create the environment needed to test CAV technology.
Issues versus opportunities
On first blush, the changes presented by driverless cars arriving on the market are complex and are likely to result in more coverage disputes (which may be costly). Insurers should remind themselves that coverage will continue to be required for physical damage to vehicles from vandalism, theft and natural causes. Further, new types of claim will need to be guarded against by consumers and, in turn, new areas for coverage will open up.
Software and cyber
Damage arising from failures in computer software is likely to become much more common. While this presents opportunities for insurers, it may also make insurers nervous. Software can, after all, be highly complex and unpredictable. Human error may also result in flaws in coding and algorithms. Since much of the relevant technology is self-learning, it could behave in ways that manufacturers and insurers alike had not anticipated. Even if insurers were able to access and fully understand the algorithms behind the software, the scope for things to go wrong is vast. The media's speculation that, with the dawn of automation, there may no longer be unforeseen accidents and insurable events seems, therefore, very unlikely. More worrying still, insurers could be exposed to large scale aggregation risk if a failure affected many vehicles.
The risk of cybercrime against driverless cars is also very real. There are research-supported concerns that terrorists and criminals may be able to engineer crashes remotely, disable a vehicle remotely, or disable elements of the vehicle's system – even from a number of miles away. Researchers discovered in 2015, that Chrysler cars, for example, could be controlled by text messages, and the manufacturer was forced to recall a number of vehicles as a result of the finding. Manufacturers are likely to seek insurance for the risks of cyber-attack, data privacy breaches and hacking. Actuaries and the insurance industry generally will need to be ready to rise to the challenge of developing models to quantify these rapidly-evolving risks (as the sophistication of the technology escalates, so too will the threats putting the tech at risk). Specialist cyber brokers will be key for understanding and advising on the nuances of policies in the retail automotive sphere. Clients will expect the insurance industry to work closely with them in order to be fully cognizant of their specific coverage needs and to ensure that coverage is flexible enough to allow them to manage their risks (see our article for more on CAVs and cybersecurity). The market has already begun to recognise the opportunity and has responded to make a wide offering of cover available.
Interaction of two worlds
There will be a long transition period between driverless cars arriving on the market and all cars on the road being fully autonomous. While one day driverless cars may be able to interact with each other in real-time, we will face a potentially extensive and unstable period of overlap where both driverless and manual cars co-exist on the road. A 2015 study found that driverless cars are more likely to be in accidents, but the accidents reported in the study all resulted from human fault in other cars which were not driverless. Even in a world where all cars are fully automated, other unpredictable (and un-programmable) road users such as cyclists and pedestrians, will still exist and present potential hazards as they interact with driverless vehicles.
During this period of overlap, insurers may try to rely more heavily on telematics data from "black box" devices to monitor driver activity. Data may include the number of miles driven, hazards recorded, and whether autonomous features have been overridden. "Dash cam" footage will also play an important role in helping to ascertain liability. Telematics data may also assist in reducing fraud by using information recorded by the car to help to root out and disprove fraudulent claims (presently a significant problem in the industry). The ABI has called for autonomous vehicles to capture data (such as whether an autonomous mode was activated and the location of the vehicle) to be made available in the event of a collision, including 30 seconds before, and 15 seconds after, an accident.
As telematics technology develops, the insurance industry will become better equipped to predict risk with more accuracy, and thus price more accurately. At present, there is very little data for actuaries to analyse in order to assess and price the risk of accidents and malfunctions. This in turn may hike up premiums. As the GAD has pointed out, weight will necessarily need to be placed on actuaries' skills of analysing data, evaluating risk and communicating information to those without specialist knowledge.
However, reliance on telematics is likely to present a whole host of new challenges for policy makers and holders alike, not least around important data protection issues. Much of the data collected from CAVs (connected and autonomous vehicles) is likely to be personal data. In the EU and UK, this means it will be subject to the General Data Protection Regulation 2016, which has strict rules on how personal data may be collected, what it can be used for and to whom it may be transferred. There may be a potentially complex relationship between the individuals whose data is being collected, the initial data controller (likely to be the vehicle manufacturer), and the insurers, which will require contractual as well as practical solutions. Some insurers are already offering drivers a reduction in premiums if they agree to allow personal data including their driving habits, to be collected directly by them (see our article for more on privacy issues).
The market is likely to want to share anonymised data for statistical analysis. Ideally, this data would be easily accessed by insurers, would be in a standardised form, and would be governed by a legal framework that ensures such sharing take place. The House of Lords has in its March 2017 report, Connected and Autonomous Vehicles: The Future?, recommended that the Government set up a Robotics and Autonomous System (RAS) Leadership Council to develop strategy regarding issues such as ethics, data protection and cybersecurity, as well as to ensure that this strategy and any data is shared across different sectors. Insurers will need to be able to access data, but manufacturers may be reluctant to produce it or may dispute certain findings – in turn, this may lead to complex and/or protracted litigation (see for example, the Nissan v Quashqai case as reported by the BBC).
On balance, there are plenty of opportunities for insurers, so long as they continue to move with the times and adapt before their competitors do. The landscape is changing so quickly and as legislation does its best to play catch up, this could be an opportunity for innovative insurers to be at the forefront of developing and writing new policies.
Teaming with technology and the importance of being dynamic
Sensibly, insurance companies continue to align themselves with (or invest in) the innovative players in the technology industry and with those who are not only developing but researching new kit. For example, XL Catlin teamed up with Oxbotica, an Oxford-based technology company, to research possible autonomous controls in a variety of vehicles beyond cars. Across the pond, Allstate Corp worked alongside data scientists and technology experts and has invested millions in researching new products and services.
New data is being produced at breakneck speeds and insurers will need to try and keep on top of it in order to analyse it. This may mean improving data analytics and recognising that existing databases will become redundant. There will, of course, also be issues regarding data protection and privacy. Since the landscape is unknown and changing each day, most insurers would baulk at trying to assess how safe an autonomous car is without the data to back it up. Some autonomous car developers have started to self-insure their own products (Google and Volvo among them) to plug the gap, or (like Tesla) have indicated their desire to offer this as part of a single package to also include maintenance. There is an argument against this that, in doing so, funds and resource will be directed away from innovation (which, after all, is where these companies' strengths actually lie). With competition getting ever more ferocious, motor insurers will need to ensure they sustain their position in the market and look to offer innovative new products. Actuaries should look to align themselves with those with their fingers on technology's pulse in order to help accurately analyse the relevant risks.
Changes in the industry may encourage more insurers to consider diversification in order to open up new revenue streams, moving, for instance, into home insurance or insurance for other technology products. We may perhaps see more price comparison sites and consumer financing. Once again, understanding where and how technology is developing and demonstrating a willingness to move with the times will be vital to staying ahead of the game. Diversifying will also be important for insurers so that they can react to new entrants into the market who will inevitably seek to disrupt and take advantage of the changing market. Flexibility and dynamism will be needed to match the technological shifts and growing complexity of those shifts.
Consumers too will want reassurance and guidance as they seek out insurance coverage. If insurers and brokers are prepared, they will be able to offer consumers the clarity they are seeking. Comparethemarket.com has recently launched the UK's first beta driverless insurance page, designed to assist customers in understanding the concept of autonomous vehicles. Of course, bringing the required clarity to consumers may have unattractive cost implications. Insurers should, therefore, try to identify potential costs so that they can be managed and kept to a minimum.
Summary of practical considerations
David Williams, technical director at AXA was quoted by the Financial times last year as saying that "We don't want the insurance industry to be blocking [driverless cars] […] By being involved early we're getting an early view of the technology". It is likely too late for insurers to think of themselves as early adopters when it comes to responding to driverless technology, but it is not too late to nonetheless seek to turn a commercial risk into an opportunity by:
- continuing to align themselves with innovators in the technology industries;
- learning to evolve and adapt and being prepared to keep doing so;
- where possible, using the data collected by cars to better anticipate and analyse drivers' behaviour and being prepared to share data within the industry (subject to data privacy obligations);
- staying ahead of the game and anticipating potential changes in how people may drive and how their vehicles may be functioning; and
- considering diversification.
The autonomous car industry may have its foot firmly on the pedal, but it will be some time before we live in a world of truly autonomous cars. The car industry sees a large number of recalls for far more simple technology such as air bags, never mind highly complex software components. It is also likely to be a while before there is a significant volume of consumers who are willing to buy such cars. Consumer confidence in driverless cars remains relatively low, with 44% of people in Britain opposing their introduction. Some manufacturers may also be wary of the increased reputational risk that would be attached to any serious incident involving a driverless incident, though with more manufacturers confirming that they would be willing to stand behind their products and accept liability for defects, perhaps any nervousness on their part may soon be a thing of the past.
Nonetheless, we are accelerating towards an age of driverless vehicles and must be prepared for the challenges that they will bring. As the GAD has pointed out, under the new legislation it will be open to insurers to develop products that they consider best fit the market. If insurers view this as a challenge rather than a chore, the opportunities will be numerous. Our view is that if insurers recognise and harness the business opportunities that the changes bring, they will be the ones behind the wheel of a reshaped insurance landscape.