The next steps in applying telematics to auto insurance pricing could involve using your smartphone to track your driving behavior, an insurance actuary said recently.
During a presentation at a December meeting of the Casualty Actuaries of Greater New York, Marty Epstein, global auto insurance actuary for AIG, revealed that smartphones are being piloted as potential replacements to dedicated boxes that collect and transmit driving data to insurers to qualify customers for discounts on their auto premiums.
Executive SummaryPeering into the future of telematics applications for insurance, an actuary envisions smartphones replacing “Snapshot”-like devices and psychological profiling of off-road behavior on the horizon.
“The big change that’s coming is this,” Epstein said, holding up his smartphone before the crowd. “This is your new telematics device,” he said, noting that the use of a smartphone in place of devices like Progressive’s Snapshot, that are currently plugged into the onboard diagnostic port under auto dashboards will translate into cost savings for insurers—and potentially higher discounts for consumers.
Epstein, who previously led a team of actuaries involved in developing Allstate’s DriveWise product, began his discussion of telematics and usage-based auto insurance by defining telematics broadly as “the technology of collecting, sending, receiving or storing information via telecommunications devices.”
During his presentation, he also revealed another trend to watch for the future—university studies that are mapping psychological profiles to driving data and providing a means to explain potential customer reactions to situations they face outside their vehicles. Drivers who frequently stomp on their brake pedals or weave through traffic at high speeds may not be careful in non-driving contexts.
If insurers know you’re an aggressive person, they may not want to offer life insurance to you, Epstein said, noting that the human behaviors that can be measured include aggression and calmness, focus, consistency, thrill seeking, avoidance, and pacing.
Marty Epstein, Global Auto Insurance Actuary, AIG
Such information might even be able to tell an insurer how responsible you are as a homeowner—information that could impact homeowners insurance pricing, he noted.
“This could be a really exciting confluence of sociology and psychology with insurance,” he said.
Although Epstein was enthusiastic about both of the trends he identified, he also listed a host of obstacles that are holding back the use of telematics in insurance.
A big barrier to entry into the user-based insurance space is the cost of the telematics devices. In addition to that, he said there is the operational complexity involved with making sure the devices get sent to customers and that the insurer is equipped to troubleshoot customer problems if they don’t work properly.
Epstein said that although prices are coming down, boxes costs roughly $100-$150 on average. Spreading that over the lifetime of the box, which could be five years, he estimated a $20 cost for the insurer. That does not include the cost of a data plan the insurer must purchase from a cellular carrier to transmit the data from the device to the cloud where data is churned to produce a risk score and determine customer premium discounts.
Adding $10-$20 per year for the data plan and another $10 for backend systems to support the usage-based insurance program, Epstein said that “before long, you’ve basically eaten up all your profit margin,” suggesting that this may be why insurers like Progressive and Allstate cap maximum discounts at 30 percent and State Farm, 50 percent.
“From what I see, for the people who are the best drivers, loss ratios could be as low as 25 percent, but nobody is giving out 75 percent discounts because there is so much operational cost,” he said.
Substituting smartphones for onboard devices could go a long way to resolving the cost issues, but would create some others, he said.
Marty Epstein, Global Auto Insurance Actuary, AIG
On the plus side, the insurer won’t have to buy a separate device and send it to the customer, and the insurer can piggyback on the customer’s smartphone data plan.
Complications are introduced, however, if there are multiple drivers (since the smartphone alternative would involve tracking the phone owner) or when the phone owner powers down the phone. In addition, he said, telematics applications are “energy expensive,” in that they drain a lot of battery power.
Turning to other factors preventing widespread adoption of telematics in automobile insurance, Epstein contrasted the volume and complexity of telematics data to the data that feeds current ratemaking variables. Under a current rating plan, the only variables that change from one year to the next might be that the driver and vehicle get a year older. Telematics devices capture hour-by-hour information about the driver, he said.
Epstein also noted that, contrary to assertions by some consultants, he believes there will be overlap between current rating variables and the new telematics information in terms of predictive value. “We would all be pretty [bad] actuaries if we hadn’t done something right up to now,” he said.
In particular, Epstein sees a big overlap with credit scores. “Maybe if we use telematics, we won’t need credit anymore,” he suggested.
Beyond the implementation problems of incorporating the new data into current pricing plans, Epstein highlighted intellectual property issues and privacy issues as additional obstacles to widespread use of telematics in property/casualty insurance.
“Progressive has been very aggressive in defending patents around the telematics space. They are involved in lawsuits with a number of carriers right now,” he reported.
Separately, in late December last year, Progressive announced that it would begin licensing the intellectual property it owns to companies interested in implementing usage-based insurance programs.
Turning to privacy concerns, Epstein noted that telematics and the related potential for behavioral research open up “a Pandora’s box of nearly unlimited intimate data.”
“Think about Google and Facebook, all the data they’re collecting on you and how little we know about how they’re using that information. Insurance companies are going to be in the same position with telematics.”
Epstein said his experience has been that insurers are treading lightly rather than exploiting the richness of the data, fearing Big Brother accusations—or worse yet, being the first defendant in a lawsuit for violating someone’s privacy.
An actuary in the audience imagined different legal issues. “Would this information be available to law enforcement? Could you just get a ticket in the mail for speeding?” he asked.
“Would this be available for plaintiffs, prosecutors and defendants to subpoena to demonstrate sobriety, drunken driving or contributory negligence in accidents?”
Epstein offered a different set of questions for the actuaries to consider:
• Will the addition of telematics to loss cost models result in significant or marginal lift in predictive power?
• Does its use bring us too close to individual risk rating?
• Can we have a credit-only model that’s going to be financially sustainable in the future?
Expanding on the second question, he asked, “What if the company knows you are driving 35,000 miles a year? Could they charge you a premium for driving three-times the average?”
On the other hand, at another point in his presentation, Epstein noted that pay-as-you-drive programs offer a way for low income uninsured drivers who don’t drive a lot to buy insurance.
Explaining the last question concerning the credit-only model, he said that telematics data is used in the United States right now only to provide premium discounts. “No one is giving a surcharge, [but] models indicate that many drivers ought to receive one for their riskier-than-expected driving.”
As to the first question, while Epstein is believes telematics data have “highly predictive power,” he concedes that other actuaries and insurance professionals “don’t think it’s going to be worth all that much.” Getting the true answer will be hard since information on telematics programs is currently highly guarded by insurers that use them, he said.
Not all insurers have been totally secretive, however. In July last year, Progressive reported findings from a detailed analysis of 5 billion driving miles, concluding that:
• Actual driving behavior is the leading variable in predicting a driver’s risk, carrying more than twice the predictive power as the second most powerful variable, which is driving record points.
• Drivers with the highest-risk driving behavior have loss costs that are approximately 2.5 times the costs of drivers with the lowest-risk behavior—suggesting that the range of rates can be far wider and more personalized than they are today.
• The majority of drivers who are lower-risk subsidize the minority of higher-risk drivers.
As consumers, the actuaries in the audience at the CAGNY meeting showed less enthusiasm than Epstein about the use of telematics in insurance, with only two out of 200 indicating that they had tried a user-based option for their personal auto insurance.
Still Epstein said he sees future customer and insurer values beyond premium discounts and competitive advantages in the market, highlighting the potential for enhanced customer interaction among them. For example, if telematics data is collected on a smartphone, then after the customer finishes a trip, the insurer could provide feedback pointing out spots along the route where the insured might have engaged in safer driving behavior.
“It’s going to be like the insurance company talking to you every day. Right now, insurers are in the background,” he said, suggesting that by increasing interaction in a way that helps customers, insurers might foster greater trust, and that trust might fuel more product sales from the same company.
Epstein noted that some companies like State Farm, make additional telematics-based customer services available for a monthly fee. These include vehicle tracking for stolen cars, vehicle diagnostic reports, monitoring of high-risk drivers in a household (teenagers and seniors, for example) and emergency-response notification (triggered by the g-force of the vehicle involved in an accident).
The Next Wave
Such services could be expanded to notify the claims department immediately after an accident, he said, noting that accident reconstruction might even be possible in the future.
“I haven’t seen too much of this in the U.S. space yet, but it is coming. These devices can tell you a lot about a claim,” he said, noting that accelerometers in telematics devices or in phones can give a good idea of where a vehicle was hit.
Speculating further on future developments, Epstein suggested that telematics devices could give feedback on your carbon footprint, tying “eco-scores” to efficient and inefficient brake- and gas-pedal pumping habits.
“Gamefication” is another possible future development, Epstein said, describing games on social networks with prizes awarded by insurers for the best drivers.
Epstein’s crystal ball also reveals potential changes in the way the data is captured for insurance pricing. He said that most usage-based insurance programs to date have been simply counting driver mistakes, such as hard-braking incidents, basing his assessment on his own prior experience working at several auto insurance carriers. “The next wave—Telematics 2.0—will take in the context of the event,” he predicted.
“If you’re braking too hard, we want to know more information.” Where did it occur? Was it on a city street? In heavy traffic of low traffic? How was the weather?
There may be a good reason for frequent braking in New York City, as opposed to California, for example, he suggested.
While admittedly more bullish than other experts about the possibilities for using telematics in the insurance industry, Epstein acknowledged that the adoption of telematics technology for any purpose has been slow in the United States. “I don’t see the penetration of telematics products getting to a critical point for quite a ways out,” he said, identifying device cost as a critical reason.
In other countries, such as Italy, South Africa and Brazil, adoption has been better with high theft rates fueling the use of telematics in vehicle recovery systems, Epstein said.