An Insurance Score is to Personal Auto as (BLANK) is to Homeowners’

If credit scores do play a statistically/actuarially valid role in determining the risk of a given driver, is there some other behavioral indicator that would play a similarly valid role for assessing the risk of insuring a specific homeowner’s home?

(Image credit: Thirsty Turf Irrigation/Unsplash.)

For many years, personal auto insurers have been using credit-based insurance scores as inputs to pricing algorithms. The insurers state that the resulting premiums are more accurate, and are thus more fair, than premiums calculated without insurance scores. In 2007 the FTC agreed.

Nonetheless, there is something very puzzling about the use of insurance scores in auto pricing algorithms, namely the lack of any obvious or intuitive relationship between

  • On the one hand, having a good credit-based insurance score
  • And on the other hand, and having fewer claims

Why should this relationship exist?

Looking at how FICO, the mostly widely used credit score is calculated provides some clues . The two largest factors in calculating FICO scores are:

  • Whether payments are being made on time (payment history)
  • The degree to which available source of credit sources are being utilized (for example using 15 percent vs. 85 percent of a credit card limit; or having one mortgage with a low balance vs. having first and second mortgages both with high balances).

One plausible explanation: the kinds of people who make debt payments on time, and who are careful regarding the amount of debt they carry; may also be the kinds of people who are more careful and aware when operating a motor vehicle. (A personal note: when I taught my son to drive, I emphasized a driver’s ethical responsibilities—telling him several times that a carelessly driven automobile could be a lethal weapon.).

So, let’s return to the question in the title of this article. If credit scores do play a statistically/actuarially valid role in determining the risk of a given driver, is there some other behavioral indicator that would play a similarly valid role for assessing the risk of insuring a specific homeowner’s home?

My hypothesis is that there is such a behavioral indicator for homeowners’ insurance, namely: The installation of devices indicating the presence of hazards that could lead to losses.

Examples include leak detectors detecting the presence of water that can damage flooring; or motion detectors indicating the presence of an intruder.

An insurer could offer discounts to a homeowner who simply affirms the presence of protective devices such as smoke detectors and burglar alarms. Or an insurer could offer discounts on a nationally recognized security systems, without receiving any data from installed systems.

State Farm  goes further. It has a program that provides free ADT home security equipment and a discount of up to 6 percent on homeowners insurance. This program does require the sharing of certain data from the ADT equipment with State Farm, in order to receive these discounts. Such fully connected programs have the potential of completing a full connected home and connected insurer cycle shown below.

(Source: Celent. Click to enlarge.)

To extend my hypothesis: homeowners who install such sensors and alarm systems, are more likely to also take steps that can mitigate other hazards in and around their homes, for example:

  • Creating a defensible space around a home in areas subject to wildfires.
  • Keeping roofs in good condition in areas exposed to windstorms.
  • Maintaining older sidewalks and driveways in good repair.

Credit and insurance scores typically fall into a broad range, for example 300 to 850—so an actuary can fine tune how to use a specific score into an auto insurance pricing algorithm.

But to my knowledge there is no widely recognized “connected home score” that provides a numeric range reflecting the number and working condition of sensors. However, it should be possible to create a connected home step function. For example crediting 10 points for simply affirming the presence of certain sensors; 30 points for providing certification of 24X7 monitoring, and 100 points for sharing data from sensors with the insurance company.

So, here is how I would fill in the (Blank) in the title of this blog:

From “An Insurance Score Is to Personal Auto as (Blank) Is to Homeowners Insurance”

To “An Insurance Score Is to Personal Auto as a Connected Home Score Is to Homeowners Insurance.”

Now it’s up to homeowner insurers to see how valid this hypothesis is. If any reader of this blog knows of a non-confidential use of this kind of connected home score, let me know.

Celent Study: Benefits of Connecting Manufacturers to Insurers

Is Connected Commercial Auto Turning the Corner?

Donald Light // Donald Light is a Director in Celent’s North America Property/Casualty Insurance Practice. His coverage areas include: technology and business strategy; transformative technologies such as digital, the Internet of Things, and driverless cars; core systems; and insurance technology M&A due diligence. His recent consulting work includes: developing a strategic IT plan for a specialty insurer, core system vendor selection support; a build vs. buy analysis for core systems; and several due diligence assignments. Light is widely quoted in the press and media, including The Wall Street Journal, The Economist, NBC and CBS Evening News, CNBC, National Public Radio. He is a frequent presenter at industry conferences including those sponsored by ACORD, PCI, and IASA.

Leave a Comment