For decades insurance companies have been searching for some device, some magic lens that would somehow tell them exactly how much risk each policyholder represents. If they knew that, insurers could price their coverage more accurately and, they say, more fairly.
Absent such a magic lens, carriers slice and dice their policyholders into broad groupings -- by age, by sex, by location, by miles driven. These groupings, experience has taught the companies, provide a good gauge of risk, at least on the average. Middle-aged woman in a small town? Low premiums. Teenage boy in an urban area? Get out your wallet.
This way of looking at policyholders works reasonably well in general, but of course drivers in these groups do vary, so that in this pricing model those at the safer end of the scale pay more and those at the riskier end pay less than their actual behavior would justify.
Is there a technological fix for this? One big insurer, Progressive Direct Group of Insurance Companies, thinks there might be, and for the past decade or so has been running experiments to try to ferret it out.
Beginning in Texas in the mid-'90s, then more recently in Minnesota, the company has invited drivers to let it hook various gizmos to their cars that would record information about their driving. What if we knew, the company's experts wondered, when, where, how fast and how far you drove? Could we really gauge a driver's risk and price it precisely?
And the answer is -- um, we need more information.
Early findings suggest there may be enough there "to augment today's pricing models," making the company "willing to continue the research effort to find out exactly what's there," said Dave Huber, product development manager at Progressive.
So Progressive is now inviting its customers nationwide to install a small device it has dubbed TripSensor on their cars, allowing it to record how often the cars are driven, when and how fast, plus data on acceleration and braking.
The device plugs into the onboard diagnostic port, which all 1996 and newer cars are equipped with, Huber said. He said the company hopes to have 15,000 cars in the program.
"If we could figure out what the biggest risk factor is," be able to say "this is the characteristic that is most likely to cause accidents," Huber said, ". . . that would be a fabulous outcome."
The program is voluntary, and participants will get $50 for providing six months' worth of data. The money is deemed a reimbursement rather than a premium discount -- as it was in Texas and Minnesota -- because a discount would require lots of regulatory filings and additional cost for the company.
Already, the company has made a couple of suggestive findings.
First, the program appears to influence driver behavior. When a motorist knows that he can save money on insurance by slowing down and cutting his mileage, it seems he's likely to do that -- potentially improving safety.
For example, speeds of 75 mph or greater create a greater risk, so time spent driving faster is something Progressive looks at, and drivers know it. "We get a lot of feedback that they are very aware if they set the cruise control [above 75] that could in some way penalize them, so they consciously set it at 74," Huber said. Or, in driving without cruise control, "their foot naturally backs off the accelerator when they reach 75," he said.
The second finding is that the current geographic rating system -- basing your rates in part on where you live -- works pretty well. On the other hand, getting detailed information about exactly where you drive doesn't help much.
The Texas program, known as Autograph, was much more technologically elaborate. The equipment put on cars included global positioning systems and cellular phones that reported not only when but exactly where vehicles were driven.
But "that turned out to be the least predictive of what I would call the where, when and how, or how much" a car is driven, Huber said. "That was a good thing because it was the scariest. People viewed that as a potential invasion of privacy. So it was okay to get rid of that."
Still to be determined are a number of nuances of driving. For example, mileage is important, and the technology captures that accurately -- an improvement from the insurer's point of view because much of the information used now is either self-reported and not terribly reliable, or derived from title records and hard to interpret.
But what isn't known is the role of driving patterns in running up a given level of mileage. Are fewer, longer trips safer or riskier than more, shorter trips?
Intuitively, you could make a case either way, Huber said. "People taking longer trips may be in a hurry or get tired. On short trips, maybe they are pulling in and out of parking lots more often," he said. The data from TripSensor could determine the reality, which in turn "allows us to identify specific behaviors indicative of risk. We can tell people about it and also price it more accurately."
Huber said drivers in both Texas and Minnesota have been happy with the programs, and most have seen reductions in their premiums. In Minnesota, drivers had the option of sending in their data every six months, and by doing so they became eligible for discounts when their policies are renewed.
More important for the rest of us, if this program works and other companies copy it, and it really does influence behavior, we may all be safer on the road.
In case you've been wondering about the priorities of people who spend time hitting a ball and then chasing it themselves, here's some more evidence about golfers. More of them know their recent score and handicap than can tell you the size of their retirement portfolio, according to a recent survey.
The poll by Nationwide Financial and the Golf Digest companies also found that 94 percent of golfers are aware of the areas of their game that need improvement, but only 45 percent know which areas of their savings and investment strategies need work.