Pricing Insurance Risk
From Henry VIII to the Age of AI
This post should probably have preceded yesterday’s post (“Insuretech: Using AI to Eliminate Bias”). So if you haven’t yet read yesterday’s post, read this one first. The topics covered will make more sense if you read them in that order. And if you can, please find 20” to watch the video embedded at the end of this post. The video does a great job explaining some difficult concepts and it is also a good example of how to do an effective presentation, a very important skill for all of us to learn.
Henry VIII was known for many things. He had six wives, each of whom suffered different fates (“divorced, beheaded, died, divorced, beheaded, survived”). His divorce from his first wife, Catherine of Aragon, triggered the separation of the Church of England from the Catholic Church in Rome during the early days of the Protestant Revolution, as well as the execution of Sir Thomas More (“A Man for All Seasons”) in a mock trial orchestrated by Thomas Cromwell (“Wolf Hall”). He also grew to be a very fat man, with a waist size reported to measure 4.5 feet in circumference. This was not good for his health or his temperament (or for the procreation of male children to succeed him, I wouldn’t imagine).
But what insurance and finance people might find most interesting about Henry VIII is that he financed one of his wars with the sale of annuities which were priced without regard to the age of the annuitant! (I don’t know if he won this particular war, but he certainly lost on the financing transaction.) Whoever bought from the King an annuity of some given amount received the same fixed annual payment for life, no matter if the annuitant’s age at the investment date was 70 or 17. That was dumb, even in an era when many people died young and few lived to ripe old ages. But in fairness to the King and his financial advisors (beheaded I should think), this transaction was done 150 years or so before Edmond Halley published his first longevity table (1693), which revolutionized the pricing of annuities and the business of life insurance.
The insurance industry has come a long way since the days of Henry VIII and Edmond Halley, benefiting from many discoveries made during the Scientific Revolution, including the theory of probability. (You simply must read Peter Bernstein’s book Against the Gods: The Remarkable Story of Risk!) Insurance companies now price lifetime annuities with reference to the age of the annuitant as well as other factors linked to longevity (gender, marital status, health conditions etc). Students from my Financial Services class will know how this works.
But most types of insurance are still priced with reference to the average risk of insured groups rather than the individual risk of the policyholder. Think about the pricing of auto insurance, for example. On average, women of a given age are lower risks than men of the same age and so women pay lower premiums for auto insurance than do men of the same age. This seems right if our frame of reference is the group (20-year old women vs 20-year old men), but it ignores significant differences in the individual risks within each group (some 20 year old women are great drivers and some are horrible drivers). And these individual intra-group risk differentials may in fact be greater than the risk differential between the group averages. (The risk for some 20-year old men is not only lower than the worst of the 20-year old women but also lower than the average of all the 20-year old women.)
But this is where much of the insurance industry is today, focused on group averages. Twenty year old women pay one premium and 20-year old men pay a different higher premium, regardless of differences in their individual risk profiles. Yes I know the industry is actually more sophisticated than this, with individual premiums adjusted for safe driver records, miles driven and other correlated characteristics (education, credit scores etc). But while the size of the underwriting groups is getting smaller (all 20-year old women vs 20-year old women with low mileage, college degrees, good jobs and high credit scores), the industry still prices its policies with reference to group averages, not individual risks.
The advent of big data and AI promises to free the insurance industry from this “tyranny of averages” and plunge us all into the era of “precision underwriting”. This is being facilitated by technological change, more and better data, and more sophisticated data analytics. And it is going to happen whether or not we are collectively ready for it. This move from underwriting group risks to underwriting individual risks promises to revolutionize the insurance industry (once we get the AI working properly), but it also raises a number of thorny regulatory and social issues that were alluded to in my last post and will be addressed more directly in a subsequent post.
But in the meantime, I encourage you to watch this 20” presentation to a group of insurance regulators by Lemonade CEO Daniel Schreiber. If you had just 20” to devote to the entire subject of insurance, I would recommend that you spend it watching this video. It really could change your life (or at least your career choices).