In October, Dan Finkelstein, a 37-year-old father of three, set out to explore buying life insurance. He went online, and to his surprise in about 20 minutes he was the owner of a $750,000 policy.
Just a year ago, Mr. Finkelstein’s purchase would have taken a month and required blood and urine samples and other medical analysis. “I was definitely surprised how easy it was,” says Mr. Finkelstein, a computer-systems architect in Ellicott City, Md.
He obtained his policy through Haven Life, a startup owned by Massachusetts Mutual Life Insurance Co. More companies, including Haven Life, are trusting algorithms—using answers provided by applicants and data pulled from prescription-drug databases, motor-vehicle records and other sources—to reveal nearly as much about many people as analysis of blood and urine.
So confident are these companies, they are making some of the industry’s best prices available for the algorithm-driven policies. Mr. Finkelstein is paying $394 a year, one of the lowest rates currently available on a $750,000 policy for a mid-30s male, according to price-comparison websites.
The firms are making a calculated bargain: Obtaining less information than before to get a deal done is better than selling nothing at all.
Sales of individual life-insurance policies have declined more than 40% since the 1980s, according to industry-funded research group Limra, and about 30% of U.S. households have no life insurance at all, up from 19% in the earlier period.
“The industry should be willing to take on this risk in order to drive a better customer experience and reverse the trend of declining life-insurance ownership,” said Mark Sayre, head of Haven Life’s policy design.
The need to go online is a sign of the times. Decades ago, life insurance was a cornerstone of families’ finances, but the proliferation of mutual funds in the 1980s provided savings alternatives. Then sales-practice scandals in the 1990s prompted many insurers to shrink agent fleets.
MetLife Inc. began online sales of policies up to $100,000 in 2011, but suspended the effort, saying improvement is needed. Numerous insurers sell small policies without medical exams, but often charge steep rates.
Mr. Sayre said Haven Life is monitoring “the experience as it emerges” to look for potential changes in evaluating applicants.
Traditionally, insurers have used blood, urine and other medical analysis to find indications of conditions that can lead to early deaths, including problems related to the kidneys, heart and liver, and diabetes.
To manage potential risks of going online, insurers say they have guardrails in place.
For starters, the firms say they can divert an online applicant back to the conventional application process if health issues surface. In addition, the electronic-only systems are mostly limited to people 45 and younger, a subset that has relatively low mortality as cancers and heart diseases typically kill older people. Insurers also typically cap the real-time policies at $1 million to limit their exposure.
Among data being tapped are prescription-drug databases, which provide access to medication histories. Insurers and their reinsurers say they have made strides designing algorithms to reflect that drugs can be prescribed for entirely different health matters. An anti-nausea drug, for instance, works for both chemotherapy patients and pregnant women, reinsurer SCOR says. Its methodology takes into account such things as an applicant’s age, overall drug profile, dosages, prescribing doctor’s specialties and medical information from other sources.
Other resources include “Risk Classifier,” a predictive-modeling tool developed for insurers by RELX Group’s LexisNexis Risk Solutions unit. It taps into motor-vehicle records; bankruptcy, criminal and other court documents; professional licenses; college-attendee records; property deeds and tax filings; liens and evictions and other records.
“Markers of a very stable lifestyle indicate a good risk,” said Elliott Wallace, who helped create the LexisNexis tool. He singled out positive indicators such as years of residence at a single address and regular credit activity.
Algorithmic-based underwriting is subject to the same state and federal oversight as conventional underwriting. Insurers must obtain permission from consumers to obtain personal information, and under the federal Fair Credit Reporting Act, databases must provide consumers opportunity to correct errors.
For younger people, motor-vehicle records are especially important because traffic fatalities are a larger proportion of overall mortality, Mr. Wallace said. High mortality rates are most strongly correlated to driving while impaired, according to research by reinsurer Hannover Re. License suspensions and revocations, and reckless driving, are next in line.
Digital data sources are far cheaper than conventional medical analysis, which can top $150 per applicant, insurers say.
As insurers migrate online, they are often partnering with web-savvy firms.
In November, Protective Life Corp. began selling term life with fintech lender Social Finance Inc. SoFi considered expanding into various types of insurance before settling on life, because, “quite frankly, the customer buying experience is the most outdated,” said Andrea Blankmeyer, vice president of finance.
Ladder Financial Inc., which is backed by the venture capitalists behind LendingClub Corp., launched real-time term-life sales Jan. 10 with Fidelity Security Life Insurance Co. and Hannover Re. Other insurers are getting in on the action through Covr Financial Technologies, an Idaho firm that provides an online life-insurance sales platform to banks and financial-advisory networks.
Covr’s advisory board includes some prominent financial-services veterans: Brady Dougan, a former chief executive of Credit Suisse Group AG; Sallie Krawcheck, co-founder of digital investment firm Ellevest, and Gary Parr, a senior executive at Apollo Global Management LLC.
Banner Life, a U.S. unit of Legal & General Group PLC, is rolling out its new digital system slowly, with human underwriters reviewing algorithmic decisions before policies are issued. The process takes about a day.
“It’s a delicate transition,” said Troy Thompson, the unit’s chief actuary, who says growing industry use of algorithm-based underwriting “may transform the business.”