Insurers flock to analytics to stretch underwriting limits

By Alain Thériault | March 28 2017 07:00AM

Karen Cutler

In recent months, insurers have stepped up their efforts to ease underwriting for advisors and clients alike. Actuaries and underwriters are teaming up with statisticians to detect high risk applicants. Exams and blood tests will soon be obsolete.

The predictive analysis of big data, also known as analytics, lets insurers facilitate insurance risk underwriting and shorten life insurance policy issue times. Insurers use public databases like Statistics Canada, or private ones, like that of MIB Group, to create a crystal ball that will let them select life insurance risks less intrusively. This may help the industry remedy its dearth of younger clients.

In April 2016, Manulife announced that it would insure eligible clients ages 30 to 65 who are HIV positive, for up to $2 million. One month later, it raised the amount of term insurance that clients aged 18 to 40 can purchase without having to supply samples of body fluids or blood from $250,000 to $1 million. The insurer extended this offer to permanent insurance in February 2017, and exempted clients ages 41 to 69 from electrocardiograms (ECG).

Manulife is not the first company to eliminate tests from underwriting. Simplified issue and guaranteed products have done this from the start. Nor is it the only insurer to drop tests for standard issue products.

In 2016, Sun Life Financial announced the possibility of covering people with HIV for up to $3 million. It also eliminated sample collection and ECGs for most insured. Sun Life says it blazed the trail three years earlier. “In 2013, Sun Life Financial Canada led the industry when we increased our testing limit for individual life insurance for clients ages 18-40 up to $500,000,” says Sharon Smith, chief underwriter and claims risk officer, Sun Life.

Sun Life continued to innovate in 2016. “In November 2016, we were the first to eliminate ECGs, stress ECGs, oral fluid samples, urine HIV tests, and medical exams for all of our life and critical illness insurance applicants, Smith says.

Automated selection platform

Launched in October 2016, Humania Assurance’s automated selection platform HuGO can accept two out of three clients without tests, for insurance amounts of up to $1 million, says vice-president, Sales & Marketing, Kim Oliphant. “Underwriting is personalized based on the client’s risk profile for maximum insured capital of $5 million. The platform accepts two out of three clients for insurance amounts of $1 million or less instantly. Most clients are permitted to purchase critical illness protection or disability protection that covers debt,” Oliphant explains.

She says 13.48% of their clients have bought complementary critical illness insurance or debt disability coverage on the HuGO platform. “Since the launch [of the platform in November], we have surpassed our objective by 250%; we are off to a flying start,” Oliphant says.

For its part, iA Financial Group will raise the limits for life insurance without samples from $200,000 to $500,000 by the end of June 2017, says Pierre Vincent, vice-president, Individual Insurance. He adds that tests and medical exams will still be required for applicants age 50 and over, for amounts of $500,000 or more.

Lee Oliphant, President and CEO of MIB Group, a joint venture that compiles medical information from insurance applications and other sources, says that insurers use the MIB databases to determine how far they can push the limits.

“Insurers can check the MIB database for self-admitted smokers or those that had previously tested positive for cotinine. As insurers strive to simplify their underwriting workflows, identifying smokers is one of the biggest questions – that information won’t necessarily be forthcoming. The use of predictive analytics, analytic underwriting engines and big data may allow insurers to skip lab tests on smoking for certain amounts of life insurance,” Oliphant explains.

Manulife confirms that it uses MIB data, and much more. “We have an analytic team that does a lot of medical research that allow us to change our underwriting guidelines. In addition to that, our analytic team develops tools and predictive analysis based on information that we collect in a number of databases, including Statistics Canada and MIB. We were able to use aggregated data analysis to predict mortality related to HIV, and come up with an insurance offer,” explains Karen Cutler, VP & chief underwriter for Manulife’s Retail and Affinity Markets.

The insurer supplements the external data with its own data. “Using years of information that our underwriting gathered from applications, we’ve also built a smoking prediction model that can identify the potential smokers. That pool of data represents half a million lives. We’ve found that the number of people that don’t disclose their smoker status is just under two per cent,” Cutler says.

Fewer smokers

Sun Life sees the smoking habit dissipating. “The number of smokers has dramatically decreased over the years, making testing for this risk less important than it has been,” Sharon Smith explains. Very few people do not reveal their whole medical history and lifestyle. “The vast majority disclose their smoking habits,” she says.

iA Financial Group also plans to rely on internal analysis when it releases a new electronic application in the first half of the year, says Pierre Vincent. “By analyzing both internal and external data, we can predict which clients should take tests,” he says.

He cites an internal analysis that found that 32 per cent of clients answered “yes” to at least one question on their insurance application. Clients who always answer “no” raise red flags. “If none of an advisor’s clients answered yes to at least one question in the application, we would have to investigate further,” he adds.

Great-West Life is hesitant to eliminate tests. “We don’t do this at this point. We continue to monitor that trend, while balancing between the need for an easier underwriting process and our needs in risk management,” says Saundra Roll, assistant VP Business Development & Solutions at Great-West, Canada Life and London Life.

These tests and exams still weigh heavily in underwriting, Roll continues. “Exams and tests help us to appropriately select the risk and price it accordingly. Results from tests and exams can sometimes reveal serious conditions that the applicants might have and don’t know about,” she explains.

Preferred rates

Traditional underwriting is the key to preferred rates, she continues. They also help manage mortality risk “which is the most important in determining the level of our dividends on our par whole life policies, Roll explains.

Another risk to keep in mind is that of client frustration. Companies that eliminate tests reserve the right to test further if the predictive information taken from applications or elsewhere signals a higher risk, Roll adds. “They may come along afterwards, and ask you to pass an exam. If they ask for an ECG or a blood test two weeks after, it can be concerning for the client,” she says.

Whatever their allegiance, insurers are keen on analytics, including Great-West. The innovation team that the insurer put in place in July 2016 runs this service. “We are using analytics for product development more and more,” Roll says.

Sharon Smith of Sun Life explains that using analytics is not new in insurance. “The Innovation Lab is made up of senior underwriters, doctors, actuaries, data scientists and data analysts. We’re also conducting medical research and working with start-ups to find opportunities for innovative diagnostics tests and designing new processes to get these tests done,” Smith says, adding that analytics lets insurers constantly refine the underwriting process.

A machine that learns

Manulife designed a similar model to underwrite risks without exams or tests. “The model we’ve built is a machine that learns as it goes, getting better and smarter,” Cutler explains. “It makes underwriting easier for people because we don’t test 100% of them. But we continue to test for higher risk conditions, like heart diseases and diabetes.”

Humania Assurance also proposes an evolving platform. “HuGO is a smart insurance platform that makes sure that only relevant questions appear,” Humania Assurance CEO Stéphane Rochon explains. How does it do this? “At first, profiling questions interact with real-time data from the MIB to generate a personalized questionnaire that is adjusted as the advisor enters the client’s answers on the form,” he explains.

Rochon adds that most policies are remitted in a few days, or even sooner. “For example, if the applicant buys a life insurance policy for $250,000 at 1 p.m., it can be approved and the policy mailed out by the insurer at 2 p.m. If all goes well, the insured will receive it the next day,” he says. Policies underwritten without tests, using Humania’s HuGO platform, can be issued within 15 to 45 minutes, Oliphant says.

Humania saves time thanks to an algorithm structured as a decision tree, which powers the HuGO platform. The insurer notes that 65% of its clients are insured in 45 minutes. Clients who opt for the instant issue process Hyperjet are insured in 15 minutes, Rochon adds. “Thanks to the smart algorithm revised every three days, I don’t need to ask all the questions. I save time. We did a test on the first 1,000 policies sold under HuGO: we saved 30 days on average. That comes to 30,000 days, or over 80 years. It’s a lot of money. It’s an efficiency gain that is almost hard to fathom, Rochon says.

Karen Cutler of Manulife also says that she significantly shortened issued times “for business in good order”.  The time to issue a policy whose application has no additional requirements is reduced from an average of 20 days to less than seven days. “In some cases, it can go as short as two days. For the business with additional requirements, we reduced the processing time from an average of 30 days to 20 days,” she says.

In the upcoming electronic application that can be used to underwrite all of its policies, iA Financial Group is adopting a form that evolves in real time. “The new application will be responsive, in that questions will gradually adjust, based on the client’s answers. Our new electronic application will now be linked to an automated pricing engine,” Vincent explains.

Co-operators Life Insurance did not have to use big data to eliminate tests, when it raised its life insurance limits without tests back in 2013, from $250,000 to $500,000, says vice president, Individual Insurance and chief actuary, Alec Blundell.

“We increased our life insurance limit for no-fluids back in June 2013, without using predictive analytics. It’s part of a strategic approach to penetrate the mid-market and their underserved needs, and effectively reach that market. We didn’t have to go to big data and predictive analytics at that time because it already made sense from an economic perspective,” Blundell explains. 

Today, however, predictive analysis is playing an important role for Co-operators. “Our business intelligence department has approximately 75 people, including statisticians doing predictive analytics. Some of them are dedicated to our life insurance line of business,” says Clément Brunet, senior director, Research and Innovation at The Co-operators.

He adds that with or without analytics, all expansions must satisfy economic criteria first and foremost. “We’re looking to a various array of other underwriting considerations, including whether to increase further our limit of insurance without tests and exams. It’s mostly about making sure clients get the best premium,” Brunet explains.

Striking a balance

Alec Blundell mentions the insurer’s reserves. “We are comfortable with a $500,000 insurance amount limit currently. Individual clients (retail) get the same standard rate as they would get if they go through underwriting. But the client still has the option to go through underwriting and get preferred rates. We are striking a balance between the individual’s benefit for good health and convenience,” he points out.

Brunet confirms the limits of analytics. He says that predictive models cannot capture all the subtleties of the traditional pricing process. In a nutshell, clients who should pay less pay more. Others who should pay more pay less.

Related to the same topic …