The Ups and Downs of Automated Life Insurance Underwriting

One of the coolest things going on in the life insurance industry is the rapid adoption of an underwriting process that seeks to streamline the process of applying for life insurance.  This so-called “automated underwriting” seeks to simplify everyone's life by using various datasets to quickly approve (or not approve) an applicant for life insurance.

We've watched several life insurers roll out, and re-roll out, this concept during the last decade.  I can say from first-hand experience that when it works correctly it's one of the most fantastic innovation launched by the life insurance industry.

But apparently not everyone shares in my love of predictive analytics, and certainly not everyone has the same faith in algorithmic decision making.

Automated Underwriting What is it?

Let's start by giving everyone a basic understanding of automated underwriting and what it so beautifully accomplishes.  To appreciate it, we have to understand the system life insurers used for decades to evaluate an applicant.

For years, when someone decided that he or she wanted to buy life insurance, he/she first sat down with a life insurance agent and filled out an application.  The application was a few (at first) and then many (as it is now) pages of particular details concerning the life insurance policy desired.  The application often included several disclosures and collected a substantial amount of medical history on the applicant.  

The agent filled out the application and collected signatures from the applicant before sending it off to the insurance company for review.

Once in receipt of the application, someone at the company typed most of the information in the application into a database system where it was later reviewed by an underwriter.

Contemporaneously, the agent (or insurance company) ordered a medical exam for the applicant.  This exam required a medically trained (nurse-like medical background) individual to meet with the applicant and collect the following information:

  • Physical measurements
  • Thorough medical background (often already collected by the agent)
  • Blood sample
  • Urine Sample
  • EKG (in some cases)
  • Saliva Sample (rarely)

Sometimes, depending on the age of the applicant or the amount of life insurance requested, the applicant needed to complete the exam with a doctor.  Some of the time, the applicant also needed a stress test performed as part of the medical exam.

Once the information from the application was in the database and results from the medical exam sent to the life insurance company, an underwriter would review the information and approve or not approve the applicant.  In many cases, the underwriter would determine he/she needed more information and would require the applicant's medical records before making a decision.

This entire process often spanned at least three months and that was usually regarded as quick in the life insurance application process.

So regular people looking to buy life insurance needed to wait (at times) months to actually buy the policy they wanted.  This caused a lot of frustration for all parties.

Enter the Automation Process

Automated underwriting isn't really new.  Several companies specializing in smaller more niche forms of life insurance provided it long before present day.  It started as an idea that for smaller death benefits, an applicant who could answer “yes” to a certain number of questions, was deemed medically healthy enough to buy the policy.  Not a terrible practice if founded on data supporting the idea that such an approach to selling life insurance didn't pose a significant risk to the life insurer.  But also not a viable option for insurers offering more robust products with substantially larger death benefits and other features that obligate the insurer financially.

So, the industry needed more sophisticated tools to offer a more streamlined application process, and those tools took shape in the “big data” world we hear about from time-to-time.  

Leveraging information compiled by data aggregators, and using advanced mathematical procedures closely related to those that created and continue to improve artificial intelligence, the life insurance industry began establishing a procedure to cut out much of the process outlined above for applicants.  

Now instead of subjecting an someone to the cumbersome process of traditional life underwriting, the insurer uses a medical questionnaire and several databases to evaluate the applicant.  If everything checks out, a decision for approval can take place well within 24 hours.  

It started with simpler life insurance products, like level term life policies.  But after a while, even the most complex whole life and universal life insurance products soon had automated underwriting options as they do today.

Automated underwriting brings quick decisions that cut the application process down to a week or so for many applicants.  As you might expect, this vastly improves the buying experience for life insurance.  It also saves lots of money associated with the underwriting process.

But just like any foray into new technology, the automated process is not without its faults and the NAIC has several tasks forces currently working on best practices to regulate and improve (I'm not sure those two words work well in the same sentence) the process.

Problems with Automated Underwriting

The biggest concern about automating the underwriting process is the risk it might pose to solvency if the insurer has a faulty approach to assessing applicants.  

If the insurer approves too many individuals that its old system might have declined, these people might pose a significant threat to the financial solvency of the company if they ended up dead sooner than the insurer anticipated.  

But solvency isn't the only concern voiced about automated underwriting.  It comes with little surprise that anytime a large company (or collection of large companies) start using large amounts of consumer data to transact business, consumer protection groups raise an eyebrow.  The practice could be entirely harmless, but there's plenty of opportunity for abuse (so they say), which leads to a lot of questions like “what information are they collecting exactly?” and “what exactly are they doing with that information?”  

At present, the life insurance industry doesn't have much in the way of specific rules for big data use/collection.  Life insurers must follow federal and state laws that govern consumer data usage, but there are plenty of gaps in those laws concerning life insurers specifically as they were not written with life insurance underwriting in mind.  

Worse yet, what happens if the life insurer's approach to this new fangled automated process results in unfair decisions for certain people?  While the life insurance industry has the prerogative to turn away certain applicants, this options comes with scrutiny especially concerning certain segments of the population.   

While it would theoretically be illegal for a life insurer to knowingly build an algorithm that turns away minority applicants (for example), there's little to no oversight presently that ensures no life insurer has done this–intentionally or otherwise.  

And what if the data used to make a decision is inaccurate?  What procedure must you follow to review the data and correct errors?  At this point, there really isn't a process to do this.

The NAIC is Working on It

The National Association of Insurance Commissioners (NAIC) has a specific working group currently reviewing a lot of information about automated underwriting.  The group is scheduled to make formal recommendations late next year concerning the regulation of automated underwriting.

There are other sub groups of the NAIC reviewing areas of automated underwriting related to bigger picture areas that intersect with it.  For example, the Life Actuarial Task Force (LATF) reviewed several areas concerning solvency and automated underwriting.  

For now we know that automated underwriting is here and I'd suggested here to stay.  There's a good chance some changes might come about as regulators suggest some changes for best practices, but those changes are probably at least a year (probably more) in the future. 

I don't think we should overlook the benefits automated underwriting offers.  Yes it might need some speed limits.  Yes we always have to worry just a little bit about abuse.  But in the meantime, we can't discount the thousands of consumers who benefit from the quick turn around that makes buying life insurance almost work like it's 2019 versus 1920.  

Leave a Comment