Concerns over price optimization are leading regulators to think carefully about the implications of longstanding practices in insurance rating, including premium “caps” on policy renewals for individuals, households and small businesses, according to one regulator.
Charles Angell, deputy commissioner and chief actuary for the Alabama Department of Insurance, delivered the news during his remarks at the 2017 Ratemaking and Product Management Workshop of the Casualty Actuarial Society, in late March in San Diego.
Price optimization refers to the use of “demand side” rating indicators that are not directly related to the potential risk of loss or cost of risk. Sophisticated, data-based price optimization is generally prohibited in personal lines, but proponents of price optimization say it is not essentially different from premium caps and other “price breaks” long permitted by regulators.
To Angell, there is a key distinction between capping premium and capping a rate. The latter is mostly not permitted, and the former allowed primarily to shield consumers from big price hikes. “If you cap a rate, your premium will clearly be inadequate,” he said. “However, capping a premium is not the same as capping the rate.”
That said, he added that capping premium based on an individual’s characteristics—e.g., propensity to shop for coverage—and not on the size of the increase “is like price optimization” and not in conformance with rating principles and laws in most states.
Furthermore, he said, even permitted premium caps can be problematic if they are implemented year after year on premiums that have already been capped. That can increase the chances of premium inadequacy.
Variable Selection Scrutinized
Angell and David Dahl, a casualty actuary for the Oregon Division of Financial Regulation, also gave examples of rating factors being flagged—and disallowed—by regulators.
It’s a maxim of statistical practice that correlation does not necessarily imply causation, and insurers often want to invoke the former without having to prove the latter to support their rating plans. There are times, however, when a risk correlation seems too far removed from reasonable causation for state insurance regulators to accept it as the basis for a rating factor. Those occasions may increase as carriers develop ever more sophisticated approaches to rating, they suggested.
For Angell, it was too much to accept when one residential rate filing included a rating factor based on the number of Social Security numbers residing at a given address over a certain number of years. The filing suggested that the more Social Security numbers residing at a location, the greater the risk.
As Angell related, it just so happened that he had allowed several relatives to live in his home over a short period and that his household might have been negatively impacted by use of such a factor in rating his homeowners insurance.
Without regard to Angell’s situation, the Alabama department found the use of such a factor to be against public policy and disallowed it.
In a similar situation in commercial lines, Oregon denied approval for a commercial trucking rating factor based on the number of relatives employees had within a certain radius. The division was concerned that the factor might interfere with a trucking firm’s hiring practices, said Dahl, in remarks following Angell.