The Death of Underwriting and Underwriters
Underwriting is not dying. It is just being automated into digital algorithms versus analog human beings.
Technology’s effect on life insurance underwriting is easiest to show by example. It is also further advanced in many ways. For example, what if consumers could take a test (which they can) that shows their probability of dying by age X? Or having results that showed the probability by which their odds of dying by age X were better or worse than average?
More specifically, let’s assume the odds of a 25-year-old dying by age 60 is 5 percent. Person A’s test shows their odds are 15 percent or 300 percent higher than normal. This makes buying life insurance for Person A, a much better investment. Alternatively, Person B’s odds are 1 percent. This makes life insurance a much worse investment.
If enough “As” take the test and buy life insurance while enough “Bs” take the test and forego life insurance, and life insurance companies do not know the test results, then sometime in the next 30 years, some life insurance companies will likely incur financial stability issues, even insolvencies. This is because they will be adversely selected against by well-informed insureds. The law of large numbers, which includes an important element of randomness, is dead in this underwriting world. All that is missing is getting enough people to take these tests, and then they buy or forego buying life insurance.
Disability and long-term care (LTC) are close second and third lines likely affected. If the odds of disease impairing one’s ability to make a living are high, one should buy disability and LTC. Otherwise, maybe an accident policy is a better investment.
On the other hand, at least one life company is offering life insurance based on an algorithm that uses downloads of patient medical data. They use that data to determine who should get offers. What are the odds that those unfortunate people are likely to die too early to get offers? Prospects offered low rates probably do not really need life insurance, at least not term, and those offered high rates probably need a policy soon, but may not be able to afford one.
Insurance Is Not Insurance
The randomness essential to insurance operating well, affordably, and as a social good, is being quietly diminished. Without randomness, insurance is not insurance. Instead, people are purchasing options, puts and calls. The only difference between stock puts and calls is who has the best data (company or client) and the best predictive model. Insurance commissioners will have much to consider regarding where these lines are drawn.
With property/casualty, consumers generally cannot take tests that give them an advantage over companies because auto, home, workers’ compensation, surety, and general liability, to name some major coverages, are mandatory purchases. But companies do have tools to specifically underwrite and rate risks that go far beyond credit scoring.
I will use an example I have not yet seen proposed. Definitive tests exist to determine a person’s tendency toward risk, risky behavior, and even ethics/crime. If I am insuring a large fleet and mandate drivers take such a test (I do not know the practicality of this, it is just a hypothetical), my pricing algorithm will be far more accurate (assuming the program is solid). The company has significantly minimized risk and the more risk is minimized, the less important insurance becomes. Companies not using the tests will be adversely selected against and will go out of business. The ones using the tests will write such perfect accounts they will barely be able to charge enough to cover expenses unless the insureds do not know they are pristine risks. At that point, actuarial pricing is not being used which is an issue in and of itself from a regulatory perspective.
Insurance by the law of large numbers created a more level field for society. The normal curve, known as the Bell Curve, plays such an important role whereby only those on the far extremes (the tails) do not have insurance, either horrible accounts that cannot get insurance or perfect accounts that do not need insurance. About 97 percent of accounts are in the middle. The change that is happening or may happen, depending on regulators, is a hollowing out of the middle. So rather than being a Bell Curve it may look more like an “M.”
In commercial lines another example involves captives and alternative risk management organizations. I understand that 52 percent of all commercial premiums are already being written in these entities. Another example is where a device becomes so safe that insurance is not necessary. Mercedes and Volvo have determined their driverless cars are so safe they do not need product liability coverage.
On the other end of the spectrum, if insurance becomes too unaffordable for a large percentage of drivers, then those that can afford auto insurance will likely need to purchase more uninsured motorist coverage and uninsured motorist premiums will likely cost more. The industry is moving towards extremes.
Fundamental Changes
When a 250-year-old industry is changing this fundamentally, adaptation will be challenging. I see and hear people talking about the future of insurance all the time. Insurance will change, and it may change so much that it ceases to be insurance.
If you are an agent, what will you then sell? If you are an insurance company, will you exist and if so, what will you be selling?
Editor’s Note: This article previously ran in Chris Burand’s column, “The Competitive Advantage,” in Insurance Journal’s print magazine on March 5, 2018.