The Race to Find the Next Insurance Credit Score (or How, Maybe, to Reinvent P/C Insurance Pricing)

What is an insurance credit score? Basically it is a set of algorithms applied to data from credit reports which provide guidance for pricing and underwriting personal lines insurance. Although it has been a source of political and regulatory controversy over the years, the use of insurance credit scores is now widespread.

Much of the controversy has been over possible disparate impacts on various societal groups. But a root cause of the controversy has been the non-intuitive relationship between a given person’s use or misuse of credit on the one hand—and that person’s probability of incurring insured losses on the other hand. It just doesn’t seem to make much sense. But statistically, there are correlations, which in general have passed regulatory review.

Insurance credit score controversies now ancient history (i.e. were settled before most millennials graduated from high school).

But suddenly something interesting is happening.

The race is on to find the next insurance credit score—and the winners (if there are winners) will gain a pricing (and underwriting) edge.

There are only two requirements to enter in this race.

1. You have to forget about all the kinds of data and information that insurers have been using to price and underwrite risks.
2. You have to use your digital imagination to find some new data and models which provide the same or better lift as the old data and models which you have just thrown out the window. (Lift is the increase in the ability of a new pricing model to distinguish between good and bad risks when compared to an existing pricing model.)

So what kind of new data might a digital imagination look at?

• For personal auto, connected cars will provide a rich data set to mine. How about whether a car is serviced at the manufacturer’s suggested intervals (correlated with whether the car is serviced by a dealer or by an independent repair shop)? Or the use of a mobile phone while the car is in motion (correlated with the time of day, precipitation, and whether satellite radio is also playing)? Or the use of headlights during daylight hours (correlated with the frequency of manually shifting gears in a vehicle with an automatic transmission).
• For homeowners insurance, connected homes could supply all types of new data. For example, whether Alexa (or other IPA) controls the home’s HVAC systems, correlated with setting security alarms before 11 pm). Or, electricity and gas consumption, correlated with use of video streaming services on weeknights. Or the number and type of connected appliances, correlated with the number of functioning smoke, CO, and moisture detectors.
• For commercial liability insurance, telematics and IoT will be the key data sources. Does a business with 10 or more commercial vehicles use both fleet management and telematics solutions? What mobile payment options are offered (correlated with dynamic pricing capabilities)? The business’ use of social media and messaging apps, correlated with the degree of supply chain digitization.

Of course obtaining a lot of this data will require permission from policyholders—and even with permission, these methods may raise social or political issues. But premium discount and loss control incentives for telematics programs have proven effective. And for better or worse, Scott McNealy got it right in 1999.