We are all used to being charged insurance premiums that are based on risk ratings. We know that we are being put into categories for health, medical, auto, or life insurance, based on some of our own characteristics. But we also know that we are being lumped in with other policyholders—some of whom might be sicker, older, or otherwise “riskier” than we are. Our insurance rates are based in part on our own attributes, but they are also based on those of the other people in our “pool.”
At times, when there is a large “pool” of insured people—for example, a pool composed of all of a company’s employees—the risk simply reflects whoever happens to be in the pool. Of course, in such a situation, we still benefit from being part of a larger pool, over which costs can be spread.
Insurance is a complex business and our premiums are calculated based on a great deal of statistical modeling. But in this column, I’ll address a simple question: Would it be better if insurance companies could calculate risk individually and thus customize a policy? Or, put another way, would it be better if we could each be a statistical pool of one?
This solution may seem appealing: Low-risk types such as better drivers seeking car insurance, or triathletes seeking health insurance, would be offered lower premiums based on their lower personal risk. This is the type of new approach that Progressive Car Insurance (“Progressive”) is now putting to the test.
Progressive’s New Approach
Progressive has developed a tool to track individual drivers’ behavior. Once the results are in, drivers will be offered insurance at a rate based on how they perform when behind the wheel. Progressive’s CEO has called this new pilot program, known as “Snapshot,” an “evolution” and “a meaningful start toward personalized insurance pricing based on real-time measurement of your driving behavior—the statistics of one.”
In this column, I will discuss the new trend towards Usage-Based Insurance (UBI), highlighting Snapshot. I will also discuss some of the privacy concerns that may arise from this type of tracking. In particular, I’ll explore how the use of GPS-tracking in the rental-car context has led policymakers to express concerns—concerns that, in turn, may lead to limits on the types of UBI that will be permitted in the marketplace.
What Is Usage-Based Insurance (UBI)?
According to the National Association of Insurance Commissioners (NAIC), UBI “is a recent innovation by auto insurers that more closely aligns driving behaviors with premium rates for auto insurance. Mileage and driving behaviors are tracked using odometer readings or in-vehicle telecommunication devices (telematics) that are usually self-installed into a special vehicle port. These telematics devices record and/or send driving data to the insurance carrier or telematics provider working on the insurer’s behalf.”
With UBI, premiums are calculated using a variety of methods, including utilizing or analyzing a customer’s activity at the gas pump, and his debit accounts, direct billing, and smart-card systems. You can tell when a person is traveling, for example, by the tolls a driver pays or how often he refuels.
Progressive was one of the first companies to offer UBI programs, about a decade ago, when Progressive and General Motors Assurance Company (GMAC) began to offer discounts for lower mileage drivers. Progressive tracked mileage through GPS and cellular systems. Several UBI models have emerged including Pay-As-You-Drive, Pay-How-You-Drive (PHYD), Pay-As-You-Go, and Distance-Based Insurance.
UBI pricing deviates greatly from that which is used for traditional car insurance—which relies on actuarial studies of aggregated historical data to produce rating factors that include a person’s driving record, credit score, personal attributes such as age, gender, and marital status, place of residence, vehicle type, driving history, and prior claims and accidents.
As NAIC explains, traditional discounts on car insurance are usually granted for having more than one policy with the same insurer, referred to as “bundling” of policies. Insurers also provide some discounts for very generic factors such as whether cars have airbags or security devices, or whether a driver has taken a driver safety course.
The rationale for UBI is that—rather than charging policyholders a fixed lump sum on a periodic basis—rates can be assessed based on more relevant, real-time data. Reportedly, there is a strong correlation between claim and loss costs and mileage driven. For this reason, existing UBI schemes seek to convert the fixed costs associated with mileage driven into variable costs that can be used in conjunction with other rating factors in the premium calculation. Imagine that your insurance might go up when you embark on a lengthy cross-country trip, but is reduced once you are back home and driving only a few miles to work each day.
UBI has the advantage of utilizing individualized current driving behaviors, rather than relying on aggregated statistics and driving records that are based on past trends. This makes insurance premiums more individualized, precise, and dynamic. Premiums can be adjusted more frequently based on a driver’s actual, recent performance behind the wheel.
Progressive and Its “Snapshot” Program
Progressive recently stated, “Historically, auto insurers have priced their products based on estimates derived from actuarial classes that include observable and verifiable characteristics like age, vehicle year/ make/model, ZIP code, claims record, etc. As good as the base science is, it lacks a direct relationship to individual driving behavior. Not surprisingly, people would prefer their rate to reflect their true behavior. They want to be treated as individuals—not as members of an actuarial class.”
Just last month, Progressive released research findings from a new five-billion-mile study relating to UBI. According to its recent press release, Progressive has been collecting real-time driving data for more than fifteen years, and has recently increased its collection efforts. Progressive boasts that in the past year alone, it has written over $1 billion in premiums that contain a usage-based premium discount.
Glenn Renwick, Progressive’s President and CEO, said, “It’s a case where the data confirms everyone’s intuition. We believed that driving behavior was the most predictive rating factor–but didn’t expect the difference to be this dramatic. Actual driving behavior predicts a driver’s risk more than twice as strongly as any other factor. It shows that we’re not members of an arbitrary actuarial class–we’re individuals with our own set of driving habits, which should be reflected in the price we pay for our insurance.”
The key findings of the Progressive UBI include the following two findings: (1) Drivers with the highest-risk driving behavior have loss costs that are approximately 2.5 times the costs of those of drivers with the lowest-risk behavior—implying that rates can be more personalized than they are today; and (2) The majority of drivers who are lower-risk currently subsidize the minority of higher-risk drivers.
Based on this study, Progressive has ramped up its advertising of a fairly recent tool called “Snapshot,” which monitors an individual consumer’s driving patterns. Progressive reports that seven out of ten drivers who try Snapshot have qualified for a discount, which can be as much as 30 percent for the lowest-risk drivers. The number of miles Progressive has analyzed has more than doubled in the last 18 months since Snapshot’s launch. Snapshot is currently available in 42 states and the District of Columbia.
Customers who want to try out Snapshot sign up and receive a small device that easily plugs into their cars. They can then log into their personal Snapshot page to track their own driving behavior and their potential savings on a daily basis. After 30 days, participants get a premium quote from Progressive that includes a personalized discount based on their actual driving behavior, which they can then compare to the rate they are currently paying with their existing insurer.
Snapshot bases a driver’s potential discount on three key driving behaviors that it tracks: (1) the time of day the car is driven (specifically, how often it’s driven between midnight and 4 a.m.), (2) the distance it’s driven, and (3) how many “hard brakes” per mile the driver makes.
The Snapshot device does not use GPS, and does not use a driver’s speed in determining the discount. The decision not to use GPS was likely a wise one, given previous regulators’ and legislators’ concerns about consumer privacy and tracking of drivers. Still, there are other concerns that are being raised about Snapshot, and about UBI generally.
Why We Should Be Cautious About UBI and Snapshot
The practice of tracking mileage and behavior information in UBI and car rental programs has raised privacy concerns. As a result, some states have enacted legislation requiring the disclosure of tracking practices and devices.
Additionally, some insurers limit the kinds of data that they collect. Progressive tracks some data that could be used in the future by other insurers, or by employers or data aggregators—and that could lead to adverse inferences about us outside of the auto insurance context.
The fact that someone drives a lot late at night, for example, may be interpreted by an employer as a sign that he is not going to stay awake or alert at work, or that he is a late-night partygoer. And while Progressive itself, as noted above, does not track our whereabouts via GPS or track whether we speed, one can surely imagine that other insurers could develop and implement tools that monitor those habits, as well as others—such as, for example, whether a driver texts and drives, or talks on a cell phone and drives.
For now, however, UBI has stayed in a less controversial area—auto insurance—where driving behavior has long been a source of our premium calculations: A traffic citation may lead to higher rates, and serious infractions such as DUIs may lead to much higher rates, and to some drivers becoming uninsurable.
But let’s take UBI a bit further: How would we feel if insurers started monitoring our daily lives, via a pedometer, thermometer, or blood pressure cuff, to determine our premiums?
Or imagine a health insurer tracking our grocery store spending patterns to see if we are making healthy choices. These data might be accurate predictors, but arguably far too intrusive to use, if they are required. But what if they aren’t required, but can be submitted by customers who want to lower their rates?
And in each context (including with Snapshot), once the data is collected, we need to ask what safeguards will stop the data from being stored and used again to rate us—in a new and different context—for example, regarding employment screening. Or maybe our current auto insurer will look at our driving data when we seek coverage for another policy—such as umbrella coverage, or homeowner’s insurance. When we submit to data collection. we can’t be sure where it will lead in the future. Of course, we should review an insurer’s privacy policies—but consumers need to be vigilant with respect to learning how their data will be used once collected.
As NAIC acknowledges, “UBI is in its infancy and thus there is still much uncertainty surrounding the selection and interpretation of driving data and how that data should be integrated into existing or new price structures to maintain profitability.”
Will Insurers’ Regulatory Requirements Prevent Abuse of UBI?
Insurers must also comply with regulatory requirements within the states in which they do business, and many states require insurers to obtain approval before using new rating methods. Rate filings to state insurance commissioners usually must include statistical data that demonstrates a connection between the data being used, and its actual connection to predicting risk.
However, a 2002 Georgia Institute of Technology survey of state insurance regulations found that the majority of states had no existing regulatory restrictions that would prevent PAYD programs.
GPS and Car Rentals: A Useful Point of Comparison
As noted above, Progressive does not track a driver’s speed, nor does it track his or her location via GPS. And the reason for this may relate to the way some courts and states have treated rental-car companies that have used GPS to track our driving habits.
The Acme and Acceleron Car Rental Cases
There are two key cases in this area. In 2001, a car renter sued an independent car rental company, Acme Rent-A-Car of New Haven, Connecticut, which was associated with American Car Rental. A renter alleged that the company, tracked customer speeds via GPS and automatically socked them with a $150 charge if they sped at over 79 mph, for at least two consecutive minutes.
Though the front page of Acme’s rental agreement included a box for renters to initial regarding tracking and the fee, the Connecticut Consumer Protection Department stepped in to stop the practice. The Department contended that the rental company failed to (1) provide renters the opportunity to refute the surcharges, and (2) notify renters that the surcharges would be automatically charged to them via direct debits or credit card charges. The Department also took issue with the rental company’s imposition of a stiff “penalty” when the company had sustained no damage.
On appeal, the Connecticut Supreme Court rightly deemed the $150 fee a penalty (and also a violation of the Connecticut Unfair Trade Practices Act). In support of its holding, the court noted this striking finding: An administrative hearing officer determined, based on the testimony of various expert witnesses, that the actual cost of excess wear-and-tear associated with driving one of the company’s cars at 80 miles per hour for two continuous minutes was closer to 37 cents, than $150.
Based on these calculations, Justice David Borden pointed out that “[t]he $150 collected by [American Car Rental] was more than 400 times the potential damage incurred.” Justice Borden also noted that “a customer would have to travel more than 1,070 miles at high speeds, without decelerating below 80 miles per hour, to [actually] cause $150 of excess wear on the vehicle.”
The Justice added that a renter who drove at 80 for two minutes, and a renter who drove at 80 for half-an-hour, would incur the very same $150 charge. The Court thus concluded that the charge did not reasonably correlate to actual vehicle wear and tear, and thus was meant to penalize the driver, not to compensate the company for harm to the rental car.
Both the trial and appellate courts in the American case declined to address the question whether American’s customers received sufficient notice of the speeding fee. The courts did not have to reach that issue, for they held that the speeding fee itself constituted an illegal penalty that was contrary to public policy. The Court stated that the issue of notice was moot because “adequate notice of an illegal penalty would not justify its enforcement.”
A second case, from California, People v. Acceleron Corp., highlights the need for companies that deploy tracking technology in car rental situations to disclose the technology’s use in a written contract signed by the customer. In that case, a car-rental company disclosed that it would charge customers $1 per mile for driving outside a designated area, but did not disclose (1) that it would use GPS to detect a car’s location, or (2) that it would impose a dollar-per-mile fee based on every mile driven, within or outside the restricted area.
Acceleron, which operates Payless Car Rental, did reference its GPS tracking, but it buried the reference in the text of its rental contract. Thus, a final settlement agreement focused on the company’s failure to adequately disclose GPS use to its customers.
According to the settlement, Acceleron agreed to “clearly and conspicuously” inform future consumers throughout the rental process, that it may use GPS or similar devices, and inform them of the manner in which Acceleron uses such devices. Acceleron was required to make these explanations in all communications with future or present customers, including at the rental counter and in a written contract signed by the consumer. One inference that can be taken from the settlement is that courts may view GPS tracking as something which is special and which requires conspicuous disclosure to drivers.
California, Connecticut, and New York Now Restrict Leasing Companies’ Ability to Use GPS
Perhaps based in part on such privacy concerns, the California and Connecticut legislatures have enacted laws restricting the ability of leasing companies to use GPS information. In California, for instance, car-rental companies may no longer use GPS information to impose surcharges, fines, or penalties relating to a driver’ s use of a rental car.
GPS information may, however, be used by a California rental company to help find a stolen, abandoned or missing vehicle. However, this practice must be clearly and conspicuously disclosed to the rental customer.
GPS information also may be used by a California car rental company “for the sole purpose of determining the date and time the vehicle is returned to the rental company, and the total mileage driven and the vehicle fuel level of the returned vehicle.” Such determinations may be made “only after the renter has returned the vehicle to the rental company.” In addition, a rental company may use GPS technology for remote locking or unlocking of the vehicle at the request of the renter.
New York has a law that also prohibits car rental companies from using GPS information “to determine or impose any costs, fees, charges or penalties on an authorized driver for such driver’s use of a rental vehicle,” subject to an exception for the recovery of “a vehicle that is lost, misplaced or stolen.”
The GPS cases and the subsequent legislation show that regulators have concerns about how companies track consumers. So while mileage is currently fair game, location tracing and mileage, taken together may not be. And legislators have already stepped in to deal with tracking in car rentals—a short-lived contractual relationship between a driver and a rental company. Imagine the scrutiny one might apply to longer-term transactions with greater impact—such as UBI insurance contracts.
Even if policymakers don’t step in, we should still be cautious about this new trend. It may feel good to get a discount on our car insurance for installing an app in our car. But will we feel the same if the app starts recording our daily habits—from whether we eat junk food in the car, to whether we frequent nightclubs, liquor stores, or dangerous neighborhoods. The sky’s the limit when it comes to how the risk factors regarding one person can be calculated, and the potential uses of data that is collected for insurance purposes may also be unlimited.
Policymakers need to be vigilant about understanding the risks involved in any use of surveillance to predict our future habits. And so do we, as consumers and customers.