States Take Aim at New Insurance Model for Setting Premiums

By Brandon Ross

Aug. 3 — State regulators and consumer groups are throwing up hurdles to an emerging algorithm-based pricing model that insurers are using to set premiums—saying the practice results in some customers being overcharged.

Just last month, Indiana became the seventh state to ban the “price optimization” model. Consumer groups and state regulators say the practice allows insurers to use statistical models to analyze individual consumer data unrelated to risk in setting premiums. As a result, the critics say, insurers can determine a customer's habits and charge more for those seen as unlikely to shop for the best price. States that have outlawed the practice say it is unfair to otherwise similar individuals and is based on factors other than actuarial risk.

“ ‘Price Optimization' is defined as any method of taking into account an individual’s or class’s willingness to pay a higher premium relative to other individuals or classes,” California Insurance Commissioner Dave Jones said earlier this year in a notice warning insurers to cease the practice. “Price Optimization does not seek to arrive at an actuarially sound estimate of the risk of loss and other future costs of a risk transfer. Therefore, any use of Price Optimization in the ratemaking/pricing process or in a rating plan is unfairly discriminatory in violation of California law.”

The industry contends the prohibitions are based on false assumptions about what insurers are actually doing. The National Association of Insurance Commissioners is expected to release a preliminary white paper before the organization's Aug. 15-18 meeting that they hope will bring some clarity and uniformity to what is allowed in setting prices. So far, California, Florida Indiana, Maryland, Ohio, Vermont and Washington have banned price optimization in rate-setting and New York has issued a letter of inquiry into the practice.

The Consumer Federation of America says more states are likely to follow suit.

The states that have banned the practice may not understand it, Robert Detlefsen, vice president of public policy at the National Association of Mutual Insurance Companies told Bloomberg BNA. “Looking at the bulletins, they don’t really define price optimization. [States] just parade this list of horrors that they assume are the result of price optimization.”

NAMIC members include State Farm, USAA, Nationwide, and Liberty Mutual Insurance.

Following Consumer Data

J. Robert Hunter, director of insurance for the Consumer Federation of America and a former Texas insurance commissioner, told Bloomberg BNA that some insurers collect all sorts of consumer data about individuals—including social media profiles, magazine subscriptions, data broker information on shopping habits and web browsing—and use that information to determine how sensitive customers are to the price of insurance policies.

“[Insurers are] moving the prices away from the risk-based or cost-based level,” Hunter said.

Vendors of the algorithm-based model, including Earnix Ltd. and Towers Watson Delaware Inc., promise in their advertisements to raise insurers profits, Hunter said. “How can they raise their profits without raising the rates?” he said. “They’re clearly raising the rates.”

Towers Watson has said that it is only selling software—the data is supplied by the insurers.

Consumer groups say such practices can be illegal because they allow insurers to consider marketplace factors, rather than just a customer's potential risk, in setting premiums. As a result, they say, individuals in similar circumstances may end up paying different rates for reasons that have nothing to do with risk.

But Lisa Brown, senior counsel and director of compliance resources for the American Insurance Association, insisted that the characterizations of how insurers are using the data they collect is inaccurate.

“No one that I’m aware of is basing your insurance rate based on how much time you’re on Facebook or which magazine subscription you have,” Brown told Bloomberg BNA.

Risk Classes

Detlefsen said price optimization is a way for insurers to “manage retention of their customers.” The model allows companies to predict a customer's sensitivity to price when buying auto or homeowner's insurance.

Detlefsen said insurers don't use the data they gather to target similarly situated individuals with different prices. Instead, he said insurers lump their customers into different risk classes and apply the optimized rates to these groups, thereby complying with the law.

Factors that insurers typically can use in grouping customers into risk categories include age, gender, car model and type, occupation, credit score, zip code and education. In theory, an insurer might only have a small number of risk classes, each with large numbers of customers. But Birny Birnbaum, executive director of the Center for Economic Justice, said it is an “incredibly misleading statement” to say that rates are applied to broad classes. He said companies have “hundreds of thousands of rating cells.”

“Many cells have one, two, or three people in them, and rarely more than that,” he said.

Hunter said there can be “thousands of classes in your tier,”

“The idea that the insurer can't almost get to the individual level isn't true,” he said. The impact, he said, tends to hit low-income consumers hardest, because they are less likely to shop around and may already have fewer insurers from which to choose in their marketplaces.

Figures Hard to Come By

Hard numbers on the number of insurance companies using price optimization and how they apply the model are something of a mystery to regulators, consumer groups and even the large insurance trade groups.

Birnbaum said the shortage of data is no accident.

“We don’t know [how many] because a lot of the insurance companies don’t disclose in their filings that they’re using it,” Birnbaum said.

Sometimes, even when insurers do indicate in state rate-compliance filings that they are using price optimization, the language of the filing makes it difficult for regulators to decipher what's going on, a Consumer Federation of America study found. If an insurer does disclose that it is using price optimization, the extent to which it may unfairly impact an individual, or not, is not detailed properly, according to the study.

Detlefsen insisted that insurers aren't doing anything shadowy to hide their use of price optimization from state regulators. States don't require insurers to detail the pricing model in that fashion in compliance filings, he said.

The Consumer Federation estimates that about half of insurers use price optimization, based on statistics released by vendors selling the algorithms, and those numbers are expected to grow.

White Paper

The white paper to be released by the National Association of Insurance Commissioners Casualty Actuarial and Statistical Task Force is expected to clarify what should be permitted. The paper is expected to be finalized around November. The American Insurance Association's Brown said the industry supports a halt to any use of price optimization to achieve illegal results.

“We all know what needs to be stopped,” she said. “There just hasn’t been a way to determine, yet, what needs to be restricted.”

Wesley Bissett, outside senior counsel for the Independent Insurance Agents & Brokers of America Inc., said most agents and brokers are concerned about price optimization and believe the issue should be scrutinized by regulators.

“It’s something that we’re focused on and increasingly concerned about,” Bissett said. “We think the reviews being done are warranted.”

CFA said in a study that Allstate has used price optimization to illegally charge similarly situated customers different rates that are not risk or cost-based. Allstate says its use of price optimization is legal, but it isn't, the CFA study said, “as several states [have] already declared.”

“They apply price optimization as a rating factor to groups of people put into Complementary Rating Groups by ‘marketplace considerations’ that are not risk-based. This method absolutely results in unfair discrimination among persons of the same risk.”

In response to the comments, Allstate told Bloomberg BNA that its prices are risk-based and fair.

“Allstate’s pricing has been and continues to be determined by risk and costs,” Justin Herndon, a member of Allstate’s Media Relations and Issues Management Team told Bloomberg BNA in a July 31 e-mail. “We use information such as driving safety record, amount and location of driving, type of vehicle and individual behaviors to provide our customers with accurate and competitive rates. We continuously update our pricing so we can make sure customers benefit from the most advanced approaches.”

“As a result, our prices are highly competitive and fair,” Herndon said.

AIA's Brown said that while there are reports of some bad actors using price optimization in an illegal way, they don't represent how most insurers utilize the collective practices that are generally known as price optimization.

“The regulators are saying there’s an egregious practice,” Brown said. “But that is not, by definition, price optimization.”

To contact the reporter on this story: Brandon Ross in Washington at

To contact the editor responsible for this story: Heather Rothman at

To see the Indiana bulletin on price optimization, go to

To see the CFA Bob Hunter Study, go to

To see a draft of the white paper, go to

To see comments on the white paper draft, go to