Car insurance telematics: why the black box should become more transparent

Fergus Cumming

Imagine you have just passed your driving test. After many hours of careful instruction, you are keen to put your good driving habits to the test on the open road. You phone up your insurance company but discover that your insurance premiums will cost you hundreds of pounds more than you can afford because “newly-qualified drivers are worse than average”. This post is about how developments in the car insurance market have the potential to revolutionise the way we drive and how we guard against the risks of bangs, scrapes and scratches. The increased use of telematics (also known as black boxes) has important implications for anyone who might consider driving, policymakers and for society as a whole.

The economics of car insurance

In some ways, insurance markets are straightforward. Buyers want to pay an upfront cost to protect themselves against the risk of large, uncertain financial obligations in the future. To know how much to charge for this premium, insurance companies want to estimate the probability of that future pay-out. But the likelihood of an incident depends on a multitude of factors, many of which are unique to the individual being insured. Some of those risks can be modelled using data known to both parties (e.g. where the car is parked or its annual mileage). There are, however, some things the driver knows that the insurance company does not.

The problem of asymmetric information has been explored by economists for over half a century. On the one hand, it can lead people behaving in suboptimal ways. So-called moral hazard is one reason why few insurance contracts offer to cover speeding fines. On the other hand, asymmetric information can lead to market closure when some participants use their private information to opt out. The problem of adverse selection is partly mitigated in the UK because third-party car insurance is mandatory, although a significant number of drivers still evade this requirement.

A combination of incomplete and asymmetric information means that insurance companies tend to rely on broad characteristics associated with groups of people, a practice sometimes known as risk pooling. Before the law changed in 2012, insurance companies would often offer different quotes to men and women. Today it is common to pay much more to insure yourself if you are under 25. To avoid being lumped into an unattractive category, a supposedly-high-risk young driver would like to level the information playing field. Leaving aside concerns to do with data privacy for the moment, telematics technology provides a novel solution.

Why my choice affects yours

A black box device fitted to a car to record information on location, speed and driving style gives our newly-qualified driver a way of demonstrating that their hours of fastidious practice were worth the effort. By showing they keep to the speed limit and accelerate smoothly they can reveal to the insurance company that they pose a lower risk to themselves and other drivers, and therefore deserve to pay a lower premium.

But my decision to reveal my driving competence might have a knock-on impact on your insurance decision as well, even if black boxes do not change driving behaviour. It is a well-known aphorism that most people believe they are above-average drivers. For these self-certified good drivers, it makes sense to opt for a device that lets them reveal how safe they really are – to put their money where their mouth is. If they are right, they should see their premiums fall relative to a world where there are no black boxes; no ability to reduce information asymmetry. Of course, if they are wrong they may end up paying more (Figures 1 and 2).

Figure 1

Figure 2

Figure 3

What happens to the premiums of those people who decide not to embrace telematics? Insurance companies assume that these people must assess their own driving practices as below average. For people on contracts without associated black boxes, their premiums should go up because the average quality of these drivers is now lower.

But what about the person who is on the fence, the person who thinks they are only a little below the average driver? They now have the choice of paying a much higher set of premiums or installing a black box and demonstrating they are better drivers than (the newly-reduced) average. They have a financial incentive to show they are better than the most dangerous drivers. The insurance company now thinks the average driving ability of the remaining non-black boxed drivers is even lower and their premiums go up again (Figures 2 and 3). As the process evolves, economic theory suggests that everyone should eventually choose an insurance contract that is linked to a black box, abstracting from other considerations that might affect preferences in the real world. We could move much closer to a world of perfect information, where premiums are tailored to the individual.

Back to reality

Universal black box adoption has two appealing features for society. First, it is fair. Instead of drivers being lumped into broad-brush buckets, they can pay a premium that closer reflects the fair price for their unique probability of making a claim. Unlike in markets such as healthcare, the individual probability of driving incidents is less determined by inherent or genetic characteristics. The most common car insurance claims are rear-end collisions, which can often be avoided with more attention paid to speed, spacing and road conditions.

Second, widespread use of black boxes is also likely to lead to higher quality driving and fewer fatalities. The WHO estimates that there are around 1.25 million annual deaths due to road traffic accidents across the globe. How many lives would be saved if drivers knew they could save large amounts of money by paying more attention to their driving habits?

Telematics insurance, however, is not without its problems. It can take time to build up an accurate profile of driving quality. Erratic driving over time is surely linked to the risk of a crash, but a violent swerve yesterday may have been the safe response to an oblivious fox crossing the road. People might also be loath to switch to black-box contracts if their no-claims-bonuses are not carried over or if they are worried about data security. Partly for these reasons, telematics insurance contracts remain a small part of the overall car insurance market.

One of the largest issues facing consumers is the lack of transparency behind the algorithms companies use to determine what makes a good and bad driver. In a report in 2017, BBC Radio 4’s Moneybox investigated several black box insurance contracts and found the reported scores awarded at the end of each journey varied significantly across time, even when tested by a professional driver in controlled conditions. Many drivers find it difficult to know which contract is right for them. If consumers are unaware of what the insurance company is looking for and the rewards to installing the black box are unreliable, the attraction of adopting a black box is severely diminished. More importantly, black-box opacity can be thought of as reversing the standard problem of asymmetric information: now the firms hold more of the cards.

Data as a scarce resource

This can be seen in a new context following the recent GDPR legislation governing the use of personal data. Since May 2018 people have had the right to access their personal data held by companies; receive a copy of it to send on to other firms; and challenge the logic of its automated processing. This provides a way for consumers to extract more of the benefits associated with using a black box.

Even when challenged, insurance firms might be loath to reveal their black box algorithms due to concerns about intellectual property. But requesting one’s own back-history of data might make insurance comparison easier if novel tools are developed to help consumers navigate the available choices with this information. This could plausibly be developed by the private sector, the regulator, or even the government. The ability to send raw data to potential new insurers could also help drivers capitalise on a history of good driving, in a similar way to portable no-claims bonuses. There is a similar discussion around the portability of another form of data that reflects good driving, in the form of uber ratings. Both possibilities could enhance market competition, making consumers better off.

GDPR has challenged us to think about personal data as a scarce resource that can be owned, sold and transferred. That has far-reaching implications beyond abating pesky marketing emails. Asymmetric information in the car insurance market is not limited to unobserved behaviour of drivers. A more transparent approach to data collection and algorithmic use would make for a better functioning market for everyone – not just those with recent memories of donning learner L-plates.

Fergus Cumming works in the Bank’s Monetary Policy Outlook Division.

If you want to get in touch, please email us at or leave a comment below.

Comments will only appear once approved by a moderator, and are only published where a full name is supplied.

Bank Underground is a blog for Bank of England staff to share views that challenge – or support – prevailing policy orthodoxies. The views expressed here are those of the authors, and are not necessarily those of the Bank of England, or its policy committees.


Filed under Insurance, Microprudential Regulation

2 responses to “Car insurance telematics: why the black box should become more transparent

  1. This totally telematic future equilibrium seems to rely on the measured characteristics being statistically significant risk factors for all drivers. Gradual acceleration and slow speeds in young drivers suggest they are not driving beyond their experience level and are less likely to have accidents, hence the lower premium. However, those same characteristics among experienced drivers might indicate a lack of confidence or a poor quality vehicle. Poor but slow drivers may be drawn to telematics and deteriorate the risk pool for established drivers, thereby destroying the discount and discouraging uptake. Good drivers with fast cars that might score poorly on telematics will also fear being penalised by installing a black box and their avoidance will leave a low-risk pool out of the system unless a noticeable discount develops. To the extent this is true, the telematics market would remain partial so long as it is an option. As self-driving technology in cars becomes widespread, the necessary data will be potentially available without installing a separate box and then providing the data may cease to be an option. That route to a telematic equilibrium seems more likely to me, albeit in a way that would take much longer than in the author’s scenario.

  2. John Watson

    Risk pooling is a positive benefit to drivers: we don’t *know* whether we are better or worse drivers than others within our risk pool.
    We do expect to have some accident claims within our lifetime. We know on average we pay the insurance company a premium over true expected return over lifetime. We are comfortable with that, because averaged within a risk pool, there is a reduction in risk even over lifetime, by averaging with others.
    But if premiums become perfect predictors of individual risk, buying fully comp insurance buys only smoothing out known liability over ones life, with high cost. There are more efficient ways of investing to achieve that.
    The rational driver would never purchase fully comp insurance.
    The insurance company will destroy their own market.