Price optimization is not a new concept, though it is receiving increasing attention in the insurance industry. Ushered in with the era of “big data” and “predictive analytics” commercial enterprises of all stripes now have the ability to refine pricing on the fly in order to extract every available ounce of economic value from their customers. At its heart, price optimization involves charging higher prices to customers who are able and willing to bear the additional cost, whether they realize it or not. Herein lies the rub.
Two customers may purchase an identical product or service but one customer pays more simply because data analytics indicated that the customer was less likely to shop around or otherwise balk at the higher price. The customer paying the higher price may not even know (probably doesn’t) that they just paid a higher price for the identical product compared to other customers. The use of price optimization on insurance premiums has all the makings of the credit-score-as-a-rating-factor controversy, if not more. Opponents of credit score premium rating argue that the credit score is not related to the cost of risk associated with insuring a particular individual – except that there is statistical evidence that it is. Circular arguments ensue.
However, price optimization is a cat of another stripe. Suppose an insured is charged more premium than would otherwise be charged simply because data analytics indicate that the customer is likely to accept and pay the premium based on the data model showing that last year’s rate increase did not cause the customer to move his business. What does that customer’s propensity to stay with the insurer in spite of a rate increase have to do with the customer’s risk of loss? Nothing, as far as I know. So price optimization boils down to charging more premium because, well, we can…. at least that’s what the predictive models tell us.
And why shouldn’t the insurance industry avail itself of all the technology and knowledge that other industries are using to their advantage? Online retailers and travel providers (especially airlines) have been doing this very same thing for quite some time. Ever heard of “yield optimization” in the airline industry? When was the last time you thought that airfares made any sense whatsoever, let alone resemble the actual cost of each passenger mile flown? So perhaps the insurance industry should use the same defense I used with my parents when I was a teenager, “Gee mom, everyone else is doing it!”
I am reluctant to get on the price optimization bandwagon, for the same reason that the “Gee mom, everyone is doing it” defense didn’t work for me a few decades ago. The frequent retort from mom was, “And if everyone jumped off a bridge, would you?” I suspect that price optimization is going to encounter greater regulatory resistance than credit scores (it already has in several states) and the cost of risk justification is far more tenuous. Even in the absence of regulatory concerns, the insurance industry might want to avoid becoming addicted to price optimization such that it begins to overshadow prudent underwriting and risk-based pricing. Imagine a world where insurance premium optimization data models become so accurate and reliable that they begin to supplant underwriting principles and the computer models start engaging in cut-throat price optimization pricing that become increasing devoid of links to the actual cost of risk. I would hope that regulators would step in before it went that far, but after 2007-2009 (and the subsequent knee-jerk reactions), I have no illusions of what regulators will and will not do.
In the end, I just have a bad feeling about “optimized” insurance premiums. The insurance industry doesn’t need another reason for consumers to dislike and mistrust the industry. I’d rather focus resources on improved cost of risk predictive modeling to better fit pricing to each insured’s risk profile, and let the airlines incur consumer wrath for playing these sorts of pricing games. Serves them right for treating us like cattle.