The traditional life insurance underwriting approach sees an agent speaking to a customer, who then fills out an application form. This is then manually underwritten, a process which might involve getting a fluid sample from the policy holder and maybe some extra medical records, and could take between four to six weeks, sometimes more.
The insurtech sector is driving significant change in this approach, fuelled by an explosion in the availability of data, the quality of data, and the technology available to analyse it.
“Taking out a life insurance policy now only needs to take a few minutes to complete online. Algorithms automatically look up third party data sources, run the results through systems and the customer gets a quote right away and they can check out instantly,” said Aaron Shapiro, CEO of US-based digital insurance company Dayforward. “The underwriting can now, in most cases, be entirely automated, which is a dramatic change to how underwriting is normally done in the US.”
According to Alby van Wyk, chief commercial officer (CCO), at Munich Re Automation Solutions (MRAS), a lot of traditional insurers have implemented automated underwriting systems, with rule engine systems that achieve good straight through acceptance rates, but he says it is important to distinguish between the US approach to life insurance underwriting and practice in the rest of the world.
“The way the US does underwriting is different compared to the rest of the world, so we look at them as two distinct market approaches. We have been deploying similar solutions into such diverse territories as Japan, Brazil, China, India, South Africa, France, and it all works because they generally have similar approaches to underwriting. Whereas in the US, insurers have upfront access to large data sets that simply aren’t available elsewhere, hence the need for a different solution,” he said.
Sydney, Australia-based Van Wyk says over a dozen different data sources are available to US life insurers which don’t exist in European or Asian markets. He cites driving records, pharmacy, and electronic health records, as examples of data points which a US life insurer has easy access to.
“Few of those data sources exist elsewhere. In the US, the focus is really about getting access to those data sources and using that data as part of the underwriting process,” he said.
Another difference between underwriting in the US and the rest of the world is the two year, non-contestability period for American life insurance contracts after they go into force. Within this time, the insurer can review the policy for fraud or misrepresentation by the policyholder. If no evidence is found by the insurer within 24 months the policy is incontestable even if the holder has lied by, for example, hiding a previous medical condition. In other countries, insurers are able to review policies at any point – even reclaiming pay-outs which are later found to be fraudulent.
“The two-year non-contestability period – and electronic access to third party data sources in North America – is largely why US life insurers do all the heavy upfront underwriting because it’s almost impossible to refuse a claim in the US after that point.” says Van Wyk.
Dayforward does use a human underwriter for complex cases or policies over a certain value. The company’s system scores each potential policy holder and produces a confidence interval for that specific rating. Too low, and the insurer gets a human underwriter to make an assessment.
“That just takes a few minutes, and the underwriter either adjusts it, and then the system improves, or they say it’s fine and a price is issued. Over time, automated underwriting gets better and better,” said Shapiro.
Whilst Shapiro believes that insurtech will produce more consistent and accurate underwriting eventually, for now, the focus for the space is aiming at achieving results that are on par with humans.
“The goal is to achieve parity with humans but underwrite instantly as opposed to a weeks-long process, and there’s no human error. So, at the moment, I would say a human could do a more accurate job on a complex case. But for more basic cases firms like ours can do a better job because we are more reliable and instantaneous.”
“A lot of insurers have implemented rules-based automated underwriting systems and are now receiving good Point of Sales straight-through-acceptance rates. This varies considerably by market but could be up to 80% in some cases. The question is, what’s next?” added van Wyk.
One answer could be AI, and van Wyk said that insurers in many markets are looking to make better use of both as part of the underwriting process.
By analysing historical data, Munich Re’s AI team observed that large numbers of cases were referred for manual underwriting, but they were ultimately classed as standard risks. By learning from historical underwriter decisions an AI model was developed, reducing the number of time-intensive and expensive requests for medical records.
“Using AI reduces the need for medical evidence, which cuts costs and improves the customer experience significantly,” van Wyk says. “There are many use cases for the use of AI in insurance, including models to stream-line underwriting, triage applications, predict early claims as well as non-disclosure.”
There may be multiple uses for insurtech to streamline an insurer’s businesses and Shapiro is confident that this points to a life insurance industry where all policies are eventually underwritten automatically.
“This is absolutely the future and I believe all the big carriers are working on projects relating to automation underwriting. Any US life insurance executive will tell you that the future of underwriting is automation. The question is: how do we get there?”