An old insurance firm bets on data with new startup
The dream of charging insurance based on the mile has been a reality for a while. But now insurers can charge based on the weather when you drive, the terrain and the type of car you’re in. The goal is to eventually do this in real time for consumers, corporate fleets and companies like Lyft or Uber.
A year ago Allstate spun out a company that could provide the underlying data for this future. Arity began with customers in the Allstate family (Esurance, Allstate) and this month scored its first customer outside the fold. National General is using data from Arity to offer personalized insurance to customers in North Carolina.
Gary Hallgren, the CEO of Arity, says that the company’s wide access to driving data and the subsequent claims data from accidents gives its algorithms an edge over data from some of the other companies offering similar data. From there it can offer quotidian analysis, such as left turns made at a specific intersection tending to cost significantly more than left turns at another intersection.
It can also offer data about specific cars that could be valuable beyond the insurance world. Halogen believes the Arity data can tell auto manufacturers about how their vehicles are performing across a wide range of conditions. Volvo would love to know that its brakes are replaced at a higher rate in cars that are sold in Montana, for example. Jiffy Lube might want to know that too.
This raises the question of how consumers will view real-time pricing for insurance data. On one hand, it’s nice to think of not paying for car insurance when you’re not on the road, but many companies already take that into account by asking questions about your commute. And as pricing gets more accurate for individuals it feels less like the insurance of old, which was defined by pools of risk.
Now the pool of risk is one. In that situation, a good driver or an infrequent driver will likely pay significantly less than someone who has to drive often, in risky geographies or even in all kinds of weather. Likely this will place a burden on those who make less money to begin with because they tend to live further from their jobs, have less choice about driving into work during a freak weather event and also have less leeway about when they come into the office. It could also penalize those who have less reliable cars.
Outside of class concerns there are positives. Granular data, if it is truly correlated with accidents, might help us reduce them simply by helping consumers or automakers avoid problematic behaviors. It could ascribe a financial cost to poor decision-making that might only have a social cost.
That becomes really interesting as the automotive industry changes. For example, as the driver of an electric car, I don’t buy gas, which means I’m not paying taxes that go to fund roads. As more and more people drive such vehicles this decreases a potential source of revenue that pays for a resource those vehicles still use. Using data on miles driven and the repercussions on roadways could become a new way to assess taxes.
In another example, a decision to drive during an ice storm could cost someone a lot more. If they were to lose their job if they didn’t go out, they likely would eat the cost, while someone simply trying to run an errand might elect to stay home. At a time when fewer drivers will mean fewer accidents, this incentive is probably good.
Finally, and Arity is actually doing this, the data it has can be used to provide safety scores for company drivers or even your Uber ride. Today an Uber star rating is almost worthless from a safety point of view. People dock drivers for smelly cars or talking too much. Imagine knowing that in addition to your driver’s ride count, you could also see that they drove well. That also could help those drivers get the insurance they need.
As a business, Arity competes with other data collection efforts including those from other insurance providers. They also will compete with companies like Waze and HERE which offer mapping and real-time traffic data.