Individuals have often been cautioned to watch what they post or otherwise divulge via social media, because employers may base hiring and firing decisions in part on what they find online. But now, there is another group that may also be watching our social networking and analyzing the data that they glean from it: insurance companies.
Has a friend posted a photo of you online in which you were smoking a cigarette? And if so, does it matter? It might, if you have previously told an insurer that you have never used tobacco. Have you said that you are a moderate drinker? Better make sure that there are no photos online of that night when you partied heavily. Are you a plumber or contractor seeking new insurance for your firm? You’d better check what consumers are saying about you online.
Such questions are becoming more relevant as insurance companies—whether they offer car, health, or homeowners’ policies—contemplate turning to social media to find out more about their policyholders’ behavior. One insurance analytics firm predicts that, within the next three years, this kind of research will begin to occur. In the interim, tests are underway to see how predictive our online data can actually be, for risk and pricing purposes.
Currently, insurance companies are turning to social-networking sites and social-media data to determine if there are incidences of fraud or misrepresentation by their customers—whether they are consumers or businesses. And recent news stories reveal that the next wave of research activity may be insurers’ using social-media information in underwriting decisions—that is, decisions about what type of risk a policyholder presents, and what rates to charge the policyholder as a result.
Data-mining firms such as Experian, Acxiom, and others already aggregate lots of data about our activities—including our online preferences and purchases. Insurance companies now represent a new market for their data and services.
In this column, I will discuss how insurers already use social media data in the insurance context, and how they may use it in the future, in the underwriting process. I will also discuss the legality of such uses—and the reasons why insurance commissioners may want to start examining this matter now.
Insurers Are Already Trawling the Web for Social Networking and Other Information If They Suspect Insurance Fraud
Even now, insurers are using information from Facebook and Twitter to nail policyholders for filing inflated or fraudulent claims.
Consider the case of Nathalie Blanchard, a 30-year-old from Quebec. Struggling with depression, in early 2008 Blanchard took a medical leave from IBM where she worked as a technician. Blanchard received monthly disability benefits from her insurer, Manulife Financial. One year later, her payments abruptly stopped. A Manulife representative informed Blanchard that Manulife had viewed photos of her on Facebook, which showed her smiling, having fun and going out to the pub—to determine that she was not depressed and able to work
Blanchard sued Manulife, accusing the insurer of failing to talk to her physician and neglecting to warn her before cutting off payments. The case is scheduled to go to trial this month.
Manulife has issued a statement claiming that it would not deny or terminate a valid claim solely based on information published on a website such as Facebook.
More generally, social-networking sites have become a new and popular tool in detecting insurance fraud, which the industry estimates costs the U.S. as much as $80 billion a year, which is reportedly between 3 and 10 percent of total annual health care spending.
Investigators who once followed people with cameras now can sit at their desks and check our Facebook pages, and/or those of our friends, to attempt to figure out if we are telling the truth. They look for things that don’t equate—like someone who claims he is too injured to work, but is pictured out dancing or running a marathon. Insurance experts caution, however, that the evidence gathered on these sites should be used only as a guidepost for further investigations and never as definitive proof of fraud.
The Next Wave: Will Insurance Underwriting Be the Next Social-Media Frontier?
What’s now up for discussion is the controversial idea of insurers’ calculating our premiums and developing our risk profiles using information from social media. This would involve insurers’ accessing a trove of data that is being assembled by giant data-collection firms. These companies sort through the details of our online and offline interactions to help categorize us as, for instance, athletes, daredevils, dieters, or couch potatoes. Moreover, when it comes to businesses such as those of plumbers or contractors, the data-collection companies may divide them into the reputable or the high-risk.
Many of us may think that there is nothing we have said or done online that would be relevant to an insurer. But we may want to think again. Are you a fan of hunting or weaponry websites? Do you post photos of yourself hang-gliding? Have you documented trips to dangerous or exotic locales? Are you a member of any organization focused on cancer or depression? Have you ever filled out an online survey in which you described your health or lifestyle? Such information, in the future, might impact your ability to get or retain many types of insurance—including umbrella, homeowner’s, automobile, or disability. It may also impact (for better or worse) the premium you are charged.
Large data aggregators are gathering online information, including from social-networking sites. Acxiom Corp., one of the biggest data-collection firms, says that it acquires a limited amount of public information from social-networking sites. The information, it says, helps clients to determine what their customers like and how their preferences compare to others, on average.
In 2010, the Wall Street Journal first reported that the U.S. subsidiary of the British insurer Aviva had teamed up with Deloitte, which looked at 60,000 recent insurance applicants and found that a new “predictive modeling” system, based partly on consumer-marketing data, was “persuasive” in its ability to replicate traditional underwriting techniques.
How Deloitte’s Approach Works: Insuring “Beth”, “Sarah,” and “Tom” and Next Generation Underwriting
Deloitte’s effort to promote predictive modeling to life insurers has gained steam in recent months, boosted partly by the Aviva research. Deloitte detailed the test at a seminar hosted by the Society of Actuaries, where a consultant helped explain Deloitte’s concept by discussing three imaginary 40-year-old insurance buyers, “Beth”, “Sarah” and “Tom”. Each had no adverse medical history and an acceptable body mass index (BMI). But data retrieved from third parties told a different story.
Using readily-available data, the experts noted, an insurer could learn that Beth commutes some 45 miles to work, frequently buys fast food, walks for exercise, watches a lot of TV, buys weight-loss equipment, and has “foreclosure/bankruptcy indicators.” “Sarah,” in contrast, was portrayed as someone who commutes only a mile to work, runs, bikes, plays tennis and does aerobics. She eats healthy food, watches little TV, and travels abroad. She is an “urban single” with a premium bank card and “good financial indicators.” Tom, the sole male example, was described as a “suburban striver” — a manager who owns his own home, is married with two kids, and is an avid outdoor enthusiast. He is also a cigar smoker.
Deloitte’s approach, the consultant report, indicates that Sarah appears to fall into a healthier risk category than Beth does. In turn, Beth seems to be a candidate for inclusion in a group with worse-than-average predicted mortality. Tom falls into an average risk pool—despite being an occasional tobacco user.
Insurance companies already use some forms of aggregated data to make decisions about us. Although it may seem surprising, insurers use our credit history and credit score to calculate premiums for us. A question such as whether or not you pay your bills might seem, at first glance, unrelated to the question of how safe you are as a driver—but insurance companies have divined otherwise.
Why Consulting Potential Insureds’ Social Networking Sites Is a Logical Next Step for Insurance Companies
Insurance companies’ consultation of social networking is, for them, a logical next step. For instance, in October, Celent—the insurance-consulting arm of the insurance-brokerage firm Marsh & McLennan—published a study entitled, “Leveraging Social Networks: An In-Depth View for Insurers,” and suggested that social-networking data could be used to help price policies.
According to the report, life insurance companies may find social media valuable for comparing what people will acknowledge about lifestyle choices and medical histories in applications, and what they actually reveal online. That could range from one person’s “liking” a cancer support group online, to another’s exhibiting some of the top signs of risky behavior—such as participation in extreme sports, or poor health and lifestyle choices.
According to various sources, individual underwriters are retrieving risk-evaluation information on their insureds through manual searches of social-networking sites. They might do this to validate an individual’s insurance form or application, for example. But in a few years, data mined automatically from social-networking sites could find its way into the underwriting process.
As users interact with multiple social-networking sites, purchase items online, and communicate with others in public forums, they leave behind data about their preferences, and habits. That data could be used to develop a risk profile for an individual, or even for a company, that changes constantly. The data can also be used to develop conclusions as to whether a policy should be renewed. Our online activity provides a fuller picture than a paper application that we fill out once a year.
According to Celent, another piece of useful information is a “social graph,” which shows how individuals or companies are linked and form networks, providing a picture of who is friends with whom, who follows whom, and who the friends of people’s friends are. This provides a picture of who we are, based on the company we keep.
Celent contends that social data has the potential to join existing third-party data sources, such as the insurance database known as “CLUE” (Comprehensive Loss Underwriting Exchange), where insurers pool claim and other information from their policyholders’ motor vehicle reports, in order to enable more accurate underwriting evaluation and pricing, and to seek to lower claims costs.
Celent predicts that insurers will turn quickly to outside data aggregators to develop tools that enable them to capture, and analyze social-media data—developing what, in essence, is a social media score. This practice will likely be similar to the process banks and credit card issuers use when they look at a FICO or credit score in order to give us a loan or extend credit.
Will the Use of Social Media in Insurance Underwriting Be Deemed Legal?
At present, insurers aren’t developing social-media scores because state insurance rate-setting regulations do not permit it. Before moving forward on that, insurance companies will need to approach coinsurance regulators with requests to use modeling based on social-networking data—but that development may not be far off.
“As with the use of credit scores in general insurance, it will take some time to determine what the boundaries are in use of this data for underwriting,” the Celent report says. One of the factors that will be used in setting those limits will be state laws and regulations.
State insurance commissioners have not yet offered official guidelines regarding the overall use of social-networking data. But they already permit the use of other third-party data in the underwriting process—including the use of credit reports and credit scores. Financial histories have thus been deemed relevant to insurance risk.
So it’s time for insurance commissioners to learn more about data aggregation and social networking: What trials are being run that are focusing on insurance and online marketing and other consumer profiles? What are the trends, and will this data prove truly relevant? Is there any benefit to consumers from the use of such data in underwriting? At present, the industry is examining this issue, and regulators, too, need to figure out the answers to some of these questions.
There are already some protective measures in place. First, insurers cannot use data that would cause them to discriminate in rates and underwriting based on race, gender, or religion. But as with credit histories, economic or financial data can be used legally by underwriters to calculate insurance premiums—and such data may often impact different groups of people disparately.
The Possible Role of the Federal Fair Credit Reporting Act
Second, and more importantly, the federal Fair Credit Reporting Act (FCRA) would likely apply to the insurance companies’ social-media analyses. If insurance companies used marketing or social media data to make underwriting decisions, that would potentially raise questions as to whether such actions would be subject to the FCRA.
The FCRA defines a consumer report as “any written, oral, or other communication of any information by a consumer reporting agency bearing on a consumer’s credit worthiness (…) credit standing, credit capacity, character, general reputation, personal characteristics, or mode of living which is used or expected to be used or collected in whole or in part for the purpose of serving as a factor in establishing the consumer’s eligibility for (…) credit or insurance to be used primarily for personal, family, or household purposes.”
If this data is subject to the FCRA, then an insurer who would take an adverse action based on this data would have to disclose it to the prospective insured. Such an adverse action under the FCRA is defined as the “denial or cancellation of, an increase in any charge for, or a reduction or other adverse or unfavorable change in the terms of coverage or amount of, any insurance, existing or applied for, in connection with the underwriting of insurance.” But this would simply mean that consumers could challenge any negative use of social-media data. It would not mean that such uses would be prohibited. Whether to prohibit such uses will be a call for state legislatures and regulators.
Celent suggests that one way to deal with this issue is for the industry to engage with customers, by first seeking permission to use their social data as part of the underwriting and application process. Depending on regulatory rules, insurers could offer customers discounts as an incentive to allow the monitoring of their social-media information.
Insurers have noted that the use of social-media data can streamline the application process for some applicants—meaning that some consumers would get a fast-track approval based on positive social-media data, while others would still go through the normal underwriting process.
Are Social Media a Reliable Yardstick for the Risk of a Potential Insured?
As regulators begin to examine the issue of companies’ using social media to calculate a potential insured’s risk, key questions arises: Are social media data really an accurate picture of a person’s lifestyle? And even if they are, are they a good measure of anyone’s risk?
For many reasons, online behavior may not accurately reflect reality. For instance, even a deeply depressed person might smile at a party, seeking to cover up the condition. Yet a claims administrator might draw an incorrect inference from the person’s photo, concluding wrongly that the person is no longer suffering from depression.
In addition to misimpressions, moreover, outright errors may arise. There is much data that is kept on us that we have no chance to see and/or correct if it is erroneous, or that may be cached before we can correct it. That’s true for social networking: How many of us have been improperly tagged in a photo? Many people have similar names: Which Joe Smith was drinking heavily at the party last weekend? What happens if I lied about my age or weight when filling out an online survey? Will another person who shares my name, and happens to actually be the age or weight I claimed to be, now be stuck with the other data I inputted about myself?
Finally, the context of posting information or photos online is very different from the context of providing information on applications or to insurers. Most people don’t worry terribly much about a typo on their Facebook page. Accordingly, if consumers are unaware that social- networking information will be used to calculate their insurance premiums, they will not necessarily provide the kind of reliable information that insurers should be taking into account to put together an accurate risk profile. For this reason, there may be very strong grounds for regulators to prohibit such uses of social-networking postings, photos, and other information by the insurance industry entirely.
Currently, insurers and data-collection firms—and the government agencies that regulate them—are only just beginning to look at the possible connections that can, or should, be made between insurance underwriting and social media. We do know, however, that insurers may examine our social-media postings in relation to a claim we file. Thus, to be safe, all of us should assume that anything we may post or tweet may be used to determine our insurability.