Blogs

Is the automobile telematics use case oversold by US insurance?

Post Comment
Senior Managing Consultant, IBM

One of the more public applications of the Internet of Things has been the insurance industry’s use of telematics, which involves putting a device in a car to record driving behavior. It captures actions such as hard braking, extremely quick acceleration, maneuvers through sharp turns, horn usage and even the car’s speed.

The insurance industry has always used approximate or correlation data such as age, gender, location and driving record to determine how much to charge for auto insurance. This is all well documented, modeled and filed with state regulators for setting auto rates. In fact, one relatively recent correlation method incorporates credit scores.

However, the use of correlation data has not always been well received. Credit scores and gender segmentation have become quite controversial both in the US and globally for being unfair and biased. Very few drivers want to have their insurance premiums determined by their gender or because they didn’t pay their student loans. If you asked consumers if they wanted an insurer to determine their auto premium based on credit score or gender, many would say it should be based on how they drive.

Insurance companies have historically spent massive amounts of time and money trying to add precision to their correlation variables. But none have seemed as promising as the use of telematics to pinpoint driving behavior.

http://www.ibmbigdatahub.com/sites/default/files/automobiletelematics_blog.jpgThe rise of telematics

One reason telematics has such potential is that because of the total number of drivers on the road today, only 20 to 25 percent are the greatest users of loss costs. By factoring in precision-based behavior, those drivers would start paying more for their insurance. Conversely, since they pay more, the remaining 75 percent would enjoy the benefit of the same precision as well based on not having to support the poorer driving habits of others.

However, being able to capture driving behavior with the aid of sensing equipment has been an affordable capability only for the last 15 years. This improvement in the technology—using data from a telematics sensing device—is thought to provide significant insight into a driver’s behavior. But if that were true, why haven’t insurance companies leveraged this data to create more accurate pricing plans? There are a few outstanding issues:

  • The industry determines pricing by driver. In a multiple-driver family, telematics cannot yet determine who is driving.
  • The driver can simply unplug the device or turn off the app to suppress the data collected—meaning the insurer is getting incomplete data.
  • While some of the context around driving behavior can be integrated to form insight, not all of it can be captured. Contextual data offers the greatest opportunity to understand the driver’s behavior.

Addressing the telematics adoption rate gap

Those three issues seem material by themselves but the largest is in the adoption rate of the telematics concept itself. While offered by many carriers, the take-up rate is not as high as it could be. Some reasons:

  • Many self-aware drivers are sure their driving habits will not entitle them to lower premiums.
  • There are drivers who value their privacy more than a discount on their auto insurance (some probable overlap with the first point).
  • The buying public generally feels insurance companies cannot be trusted.

Plus, despite years of research and analysis of an incredible amount of data, there has been insufficient development of correlating evidence to suggest that any one particular behavior increases the probability of an insurable loss. For insurance carriers, the telematics use case has been about rating accuracy and fairness to the driver. Without the necessary correlation, the actuarial models for pricing cannot be “tuned” to this information.

So, can we conclude that automobile telematics use cases are oversold? In one way, yes. Determining pricing from the current experience in the US suggests that this is not the best use case, at least not yet.

I think there are other opportunistic plays available today. However, these call for innovative thinking by the industry such as:

  • Integrating weather, telematics and GPS data to warn drivers of hazardous road conditions while their commute is in process
  • Using weather data and the approach of a significant storm to predict exposed assets for claims planning and servicing needs
  • Applying fraud detection techniques that consider vehicle position and a storm's path to anticipate any potential issues
  • For new or inexperienced drivers, providing immediate online feedback on driving behavior to point out errors or hazardous mistakes, helping to create better driving habits
  • Aggregating data to identify frequent errors or behaviors occurring at incident hot spots, enabling local municipalities to apply traffic safety and control methods (rumble strips, stop signs, metering lights and so on) to create safer roads

Discover more innovative thinking in the insurance industry and learn how analytics for the insurance industry is enabling insurers to leverage data and analytics to create a customer-focused enterprise while maintaining the required oversight to succeed.