How big data can make insurance better
BY Pierre Ghorbanian, Advisor.ca | April 6, 2017
A recent health setback has me thinking about how tech can improve our clients’ experiences with medical professionals and insurance companies. For instance, I was impressed that my general practitioner was able to electronically pull medical reports from a local walk-in clinic that she was not affiliated with – potentially helping her make a better diagnosis.
This had me thinking about how long it takes for insurers to get information from doctors and hospitals in order to process a claim or new application. In my experience, the industry is held back by legacy computer systems, which in many ways are an even bigger competitor than our peer companies.
How could insurance products change if insurers could freely upgrade their legacy systems?
What the future could be
Imagine this scenario.
You suffer a critical illness. Upon diagnosis, the hospital immediately sends the insurance company your medical information — electronically. That starts the claims process. By the time you’re discharged, the insurer has received the information needed to approve your claim, and done so.
Fast forward 40 years later, and you’ve died. The results are immediately forwarded electronically to your insurance company so it can process the death claim. Your family promptly receives the death benefit.
And the story doesn’t end there. What if your entire medical history could then be analyzed to determine genetic risk factors and inherited conditions? The insurer could then determine whether your death or your illness 40 years ago was related to genetics or inherited conditions, and create better risk models and tests for future clients.
How to use analytics
An insurer’s actuaries could use robust data analytics to find relationships between claims activity and insureds’ behaviour and characteristics. Keeping all other variables constant, they could determine how a specific incremental increase or decrease would impact mortality or morbidity. For example, does an incremental increase in public transit usage cause mortality or morbidity to decrease? Or, can an ethnic group predisposed to a genetic risk factor decrease mortality or morbidity risk with a specific environmental or lifestyle changes?
And, by using the Internet of Things, insurers could connect to people’s devices to obtain real-time data and find relationships. (Privacy issues and legislation will need to be considered if this becomes a reality.) The data could be used to improve product design, and optimize pricing as well. If there is a specific risk that increases cost, other improvements could reduce overall costs. Alternatively, other value-added benefits could be added to offset the higher cost of a specific risk (e.g., a second opinion service, free access to a private clinic in the U.S., etc.).
Now that the data analytics team has determined how to price a person’s insurance plan, what’s next? Perhaps they could refine their risk modelling and build better products to reflect the probability of getting a specific illness for someone of your demographics or psychographics. Some aspects of pricing or product benefits could improve because a specific risk is much lower than we currently assume.
Overall, analytics shouldn’t increase the probability of claims being denied, especially on fully underwritten products. I’d argue that the more information upfront means less chance of a claim being denied.
When it comes time for clients to buy insurance, there’s also room for evolution. Non-face-to-face methods (NF2F) are growing in popularity; today’s NF2F applications can include a simple fillable PDF that clients can email to their advisors, all the way to advanced mobile applications. Video chatbots and artificial intelligence could be the next way to collect information.
Beyond that, outside of North America we are seeing peer-to-peer (P2P) insurance applications. That’s where small social groups with common interests purchase non-traditional insurance solutions. Existing examples include lemonade.com, a New-York based property insurer that charges fixed instead of variable premiums (and donates unclaimed remainders to charity), and friendsurance.com, a German insurer that rebates policyholders up to 40% of their premiums if no claims are submitted.
P2P platforms can potentially attract consumers who are currently self-insuring. For example, a February 2017 study by the Bank of Korea determined that loss-averse people are less likely to own term insurance and more likely to own permanent cash value plans. The rationale was that loss-averse folks who do not make claims would feel as if their premiums paid were, in fact, losses. For them, a cash value insurance policy is a good alternative, assuming it’s affordable. But a P2P platform could help provide a better experience and utility of owning an insurance plan even if a claim was never made. For instance, if the P2P platform had a loyalty program, premium rebate or referral program, or if it used gamification, it could attract someone not interested in a cash value policy.
How will fintech innovation alter purchasing behaviour? How will products be developed and distributed in the future as a result of innovation? How will insurers adapt when the next disruptive technology enters the industry?
Insurers must think about the issues; governments are already trying to get a handle on those questions. For instance, the Ontario Securities Commission has started LaunchPad, which helps fintech players work within regulations, and helps regulators adjust as needed.
An educated guess: commoditized, transactional purchases will become automated. With that said, more complicated tax and estate planning strategies will likely still continue to depend on more personalized traditional one-on-one purchases. Everything in between will be the grey area many advisors can work within in the next decade or two.
Pierre Ghorbanian, MBA, CFP, FLMI, is the advanced markets business development director at BMO Insurance.