Car Insurers Find Tracking Devices Are a Tough Sell

January 11, 2016

As Seen On: The Wall Street Journal

MAYFIELD VILLAGE, Ohio—In a lab at Progressive Corp.’s suburban headquarters, managers and technicians at the auto-insurance giant watch a colleague drive from his home outside Cleveland to the city zoo. A dot on a computer screen moves along a map overlaid with speed limits, turning red when the driver is speeding.

The demonstration is part of an effort to figure out how location data could someday be used to assess drivers, as the roughly $185 billion auto-insurance industry undergoes a wholesale change in how companies size up risk. Using devices installed in customers’ cars, Progressive and other auto insurers are tracking things like how far their customers drive and how often they slam on the brakes.

For decades, insurers have relied heavily on lumping applicants into broad actuarial categories by characteristics such as age, gender and car type to supplement information gathered from driving records. Now, they are collecting reams of personal data in hopes of drawing much more tailored conclusions.

The catch is the industry needs its customers’ consent to gather the information, and many customers are wary.

“I know some people say, ‘What do you have to hide,’ but I don’t want big business or Big Brother being involved in my personal life,” says Shauna Aiken, a San Diego driver of a Mazda Miata. “It just creeps me out.”

At a time when industries such as advertising, media and technology are being reshaped by an explosion of personal data, the effort by auto insurers is turning into a test of the extent to which Americans are willing to share private details for potential economic benefit.
Progressive, Allstate Corp. and State Farm Mutual Automobile Insurance Co. are promoting plug-in devices and/or smartphone applications, as well as cutting deals to piggyback on auto makers’ in-car computers, all in hopes of getting sensors into cars.

For insurers committed to this new way of sizing up accident risk in policyholders, the appeal is that they can better identify safe drivers, who are less likely to file claims, and entice them with discounts off the conventional pricing—potential reductions up to about 30% are possible. The goal is to keep those less risky customers long-term, while riskier drivers facing rate increases could be motivated to shop around.

Progressive, the fourth largest U.S. auto insurer, says nearly 80% of drivers qualify for discounts. Until recently, it and other companies running these programs haven’t surcharged those drivers they learn through monitoring are worse than the traditional methodology indicated, in the interest of attracting as many participants as possible and building up databases to analyze, though Progressive in late 2014 began levying surcharges of up to 10% in some states.

Progressive has about a quarter of its new customers using its monitoring device. Allstate, which got into the game later but is ahead of Progressive in rolling out a smartphone-application, says its take-up rate is about one in three.

Winning trust
Chief Executive Glenn Renwick says Progressive recognized from the beginning that it would be an uphill battle. Early company surveys of people’s interest in usage-based insurance revealed that about 40% of people had a viewpoint that was some variation on “No way in hell.”

“Insurance is not something where people say, ‘I trust you,’ ” he says.

Progressive traces its interest in data to its entry into the car-insurance business in 1937, after other insurers had largely cornered the market for the safest drivers. Left to pick through pools of rejects, Progressive studied accident reports to find the relatively better risks. It discovered, for instance, that older people who failed to yield at intersections posed more risk than those with speeding tickets.

In the 1990s, its analysts helped establish a correlation between credit histories and claims by testing a hypothesis that financially responsible people are more cautious drivers. They factored that into pricing, so some applicants who on the surface look risky could end up paying less, and vice versa. Many rivals followed suit.

Progressive’s managers then proposed gathering information directly from cars in hopes of crafting truly individualized rates.

Over the next decade, it tested a series of devices that had to be installed or plugged into a customer’s car. The devices were too cumbersome to catch on with the broader public and too expensive for Progressive to put into widespread use.

Still, Progressive gleaned new information, such as a mile driven at 2 a.m. was in general four or five times riskier than one driven at 7 a.m. One early device contained Global Positioning System technology, but because location tracking turned off so many consumers, Progressive stripped it out.

In 2008, Progressive came out with a device that used wireless technology, which meant data could be sent directly to Progressive’s computers. It began rolling out a redesigned version in 2010, calling it “Snapshot.” The name reflected a changed tack: Instead of monitoring drivers all the time, it decided to limit its assessment period to six months.
Mr. Renwick, a former engineer at Bell Labs, says the company concluded most people have repetitive commuting patterns and driving behavior, so managers decided it would be enough to just take a snapshot, after which the device would be removed.

Since 2008, nearly four million drivers have participated in Progressive’s program. The most powerful predictor of accidents, the company says, is hard braking. Progressive defines that as a decline of 7 miles an hour in the space of a second. Progressive’s device beeps every time it senses a hard brake to make drivers aware of the risky behavior.

The conventional approach of grouping insurance applicants into demographic pools reflects findings including that the average married male is 3% more likely to have an accident in a given year than a married female, according to Progressive. But married males with the fewest hard brakes are 41% less likely to have an accident than the typical married female as determined by the conventional methodology, according to Snapshot data.

Allstate also tracks hard braking, defining it as a decrease in speed of 8 miles an hour or more in a second. The company has found that drivers who slam on brakes more than eight times every 500 miles are 73% more likely to get into an accident in a given year.

Customers who have signed up include Vanessa Irwin of Farwell, Mich. Three years ago, she was attending community college, working part time and couldn’t afford the couple hundred dollars a month she was quoted to cover a 2003 Chevrolet Impala, thanks to a ticket for speeding on a 70 mph stretch of highway. Then her grandmother saw Progressive’s Snapshot commercials.

“My grandma kinda talked me into it,” says Ms. Irwin, now 22 years old. “She said, ‘What can it hurt if it saves you money?’ ” Within a month, her rates were cut by 30% to about $165 a month.

During her six-month Snapshot test, she recalls just a handful of beeps for hard braking.

Ms. Irwin says she recommends the program but notes her friends don’t always bite. “They have privacy concerns,” she says.

The ‘bunker’
At Progressive, data on about a million new trips pour in each day, and are logged in computer servers housed in a building called the “bunker” at the company’s headquarters. The servers hold data on 15 billion miles of driving.

Progressive’s privacy policy says that Snapshot data is encrypted when transmitted to its servers. According to the policy, Progressive won’t sell personally identifiable Snapshot data to third parties, and it won’t use the data to resolve insurance claims without first obtaining permission. It also won’t share the data with any third parties “unless it’s necessary or appropriate to service your insurance policy, detect or prevent fraud, perform research, or comply with the law,” such as in response to a subpoena in a civil lawsuit or by police when investigating the cause of an accident. Progressive said it has received subpoenas in “a small number of instances,” declining to provide specifics. The data stays in Progressive’s computers after the six-month period is over.

Progressive hopes to coax more participation out of its customers when it launches a smartphone-app version of its device in 2016.

David Pratt, who runs Progressive’s Snapshot program, says smartphones have made Americans more comfortable with location tracking, so the company is once again researching GPS use, including in some of its Snapshot devices.

With the spread of smartphones lowering the barrier to entry, small and midsize insurers are plowing ahead, tapping into the large number of consulting firms, app developers and device makers available to help.

Companies that don’t offer a usage-based program could be impacted by “adverse selection,” according to Carl Smith, a director of product management in the U.S. car-insurance operations of Spanish insurer Mapfre SA, meaning they could be stuck with many bad drivers as rivals’ usage-based programs attract good ones. Mapfre began rolling out a program in 2012.

Among the few big insurers that have yet to publicly acknowledge at least experimenting with usage-based insurance is Berkshire Hathaway’s Geico unit, which has bigger market share than Progressive. In the past, Berkshire CEO Warren Buffett has said that Geico doesn’t see evidence that usage-based pricing would provide an advantage. Through a spokeswoman, he declined further comment.

State insurance regulators generally have welcomed usage-based insurance programs because they are voluntary and provide the opportunity for drivers to learn how to improve their driving. Still, location tracking raises bigger concerns, and not just for the privacy implications, some consumer advocates caution.
Some people could get higher pricing “not because of the manner of driving, but because of where and when they drive due to long commutes to work or driving through neighborhoods disfavored by insurers,” says Birny Birnbaum, a former Texas insurance regulator who now heads an Austin-based nonprofit, Center for Economic Justice, which advocates for low-income consumers.

Progressive’s Mr. Pratt says: “It’s too early to tell how location data may factor into determining rate.”

Insurers are still on a learning curve when it comes to usage-based systems. Last year for instance, Progressive concluded Snapshot was overestimating the potential for accidents from midnight to 4 a.m. driving on week nights, and now considers late-night driving to be high-risk only on weekends.

Insurers and their trade groups note that usage-based programs are subject to regulatory review by state insurance departments and must be in accordance with laws that prohibit unfair pricing practices.

To attract more users, Allstate in August announced new features for users of its usage-based “Drivewise” program. They include points for deals on merchandise, gift cards and local offers through safe-driving challenges, such as driving below 80 miles an hour and avoiding hard braking for three straight days.

Allstate, which launched its plug-in device in 2010 and the cellphone version four years later, says about 925,000 of its 20 million policyholders are enrolled in its program. It is cutting expenses elsewhere but not with Drivewise.

“We’re investing quite heavily,” Allstate CEO Thomas Wilson said.

Technological innovation is rapidly eliminating some of the hitches in the monitoring programs. State Farm has a cellphone-app program in Ohio in which it provides, free, a small, Bluetooth beacon that a policyholder puts in a glove compartment so that the app tracks driving behavior only when the policyholder is in that particular car, as opposed to a taxi, bus or someone else’s car as a passenger. Under many other cellphone apps, a policyholder has to go into the app to strike out a given trip if he or she wasn’t actually behind the wheel.

More detailed monitoring may be on the way. A patent issued to Allstate last year covers sensors and cameras that would record “potential sources of driver distraction within the vehicle (e.g. pets, phone usage, unsecured objects in vehicle).” The patent says sensors could “detect the content of alcohol in the air,” as well as a loud stereo or other noise. Sensors also could track nearby cars’ driving patterns.

Paul Saffo, a specialist in technological change who teaches forecasting classes at Stanford University, sees usage-based pricing as “probably inevitable long-term.” He says U.S. consumers have shown in recent decades a willingness to sacrifice privacy for convenience, and he believes they will continue to do so even as they complain.

“Privacy already has shifted from being a right to a good that is purchased,” he says.

Write to Leslie Scism at leslie.scism@wsj.com