DENVER — Every time you try to get an insurance quote, whether it’s for a car, home or family, companies compile a series of data to try to determine the risk each individual poses.
They use that risk factor to determine how much the monthly rate should be. The goal is to strike a balance between how much insurance companies collect with premiums versus how much they pay out in claims.
In the age of big data, companies use things like age, gender, income level, safety record, neighborhood and more to set those rates.
“We have been told that there can be up to 80 risk factors that are evaluated looking at insurance,” said Sen. Janet Buckner, D-Aurora.
However, many companies also use some unobvious data points to draw that conclusion. Chuck Bell from Consumer Reports says their data has found car insurance companies, for example, might use your job title, education, home ownership status, credit score and more to set rates.
“There has been a problem with some insurance companies even using your media usage habits to see if you like to switch your cable providers and if they think you’re not going to switch your cable company and you’re going to remain a loyal customer then they may keep your premium at a high-level,” Bell said. “Why is the insurance company even looking at that kind of data?”
The practice is called price optimization; the idea is that the less likely a consumer is to shop around, the more loyal they’re likely to be even if rates are high.
For car insurance, consumers might expect their driving record to factor heavily into their rates, however, a 2015 Consumer report study found that a person with a perfect driving record but a poor credit score will pay, on average, $1,100 more than someone with excellent credit but a DWI conviction on their record in Colorado.
Colorado lawmakers are trying to put some guardrails around the types of data insurance companies are allowed to use in setting those rates.
Senate Bill 169 would prohibit insurance companies from using data indicating that someone is in a protected class, such as their race, gender or sexual orientation, in determining premium rates.
“The issue is there is so much data and algorithms information that’s being put into a computer, basically no one knows how they’re being judged,” said Buckner, the bill’s primary co-sponsor.
Buckner believes she has found instances where there is inherent discrimination in the data. She has collected data to show that a 35-year-old man living in Park Hill, where 56% of the population is black, would pay hundreds more in car insurance rates than if that same man was living in Aspen or Boulder, where the populations are primarily white. Buckner says her data shows the man would pay nearly $140 less in Boulder and nearly $250 less in Aspen.
“Why is there such a big difference? And one of my biggest concerns is: are we subsidizing other zip codes because of this practice?” Buckner said.
She wants insurance companies to take a closer look at their own data aggregation to determine whether there is unintentional discrimination at play. Buckner says she doesn’t blame the insurance companies, but she would like to make sure data is being used appropriately and not to propel inequality.
Bell agrees with the bill and wishes it would go even further to include things like credit scores. Five other states have already banned the use of credit scores: California, Massachusetts, Hawaii, Michigan and Washington.
Other states, like New York, have banned the use of education level to set insurance rates.
“We’re concerned that insurance companies are engaged in a form of high-tech redlining, which is that they discriminate against some of their customers by using big data to create a socioeconomic profile of that person,” Bell said. “There might be hidden patterns in the algorithm that are just inherently unfair to some subset of the population and companies are going to have to become more transparent and accountable for the data sets they use.”
Others worry about the unintended consequences the bill could pose.
Kelly Campbell from the American property and casualty Insurance Association worries SB 169 would create a fundamental shift in the insurance industry that could raise some rates.
“This type of broad, quick action would have unintended consequences and hurt the people that it’s purportedly trying to protect by costing more on all types of insurance,” Campbell said. “They may no longer be paying their insurance based on the risk they present to their insurance company. This could force them to pay more than the actual risk that they present.”
One example Campbell provided deals with teenage drivers; data shows that young, female drivers are more likely to drive safe and less likely to get in a car wreck than teenage males. As such, their insurance rates might be a bit lower based on their gender.
If gender is no longer allowed to be a factor, Campbell says the insurance rate could go up for those women.
“The more information and insurance company has about an individual, it’s better to understand that risk,” Campbell said. “When we are forced to use less information than we don’t have as good of a sense of the type of risk an individual presents.”
She believes the bill is too broad, too ambiguous in parts and happening too quickly for lawmakers to understand all of the potential effects.
“Underlying that is a concern that if insurance companies can no longer rate according to an individual’s risk, can insurers collect the proper amount of premium to pay for the claims that they’re going have to pay for?” she said.
If not, she worries premiums could be raised for more people unfairly or become one-size-fits-all.
In the age of big data, SB169 would put more guardrails around the types of information insurers could use to set rates if passed. The bill passed its final reading in the Senate Thursday and now moves on to the House.