The American Property Casualty Insurance Association says California’s insurance marketplace is “fragile,” and has joined with other organizations calling for reforms to modernize the state’s rules and regulations when it comes to the homeowner insurance market.
The association was one of 30 presenters participating in a recent workshop hosted by the California Department of Insurance to gather public input on the use of catastrophe modeling in insurance rate-making.
Catastrophe modeling uses computer algorithms to assess potential losses to homes, for example, due to man-made and natural disasters.
The webinar’s presenters addressed legal questions surrounding how to comply with Proposition 103 public inspection and transparency requirements.
Prop. 103, approved by voters in 1988, outlines rules governing state insurance oversight. It requires “prior approval” by the state before insurers can enact property and casualty rates. Critics say the public does not have access to insurance company “open books” and the risk assessment rationale used when proposing rate changes.
The state asked participants in the July 13 workshop to discuss how a risk management model should be structured, as well as ways to implement it with a public inspection requirement.
California Code makes it illegal to include forward-looking climate change projections in catastrophe modeling for fire insurance rates, and insurers must estimate future losses using their average historic losses for the past 20 years. Those averages do not reflect extensive housing growth in high-risk regions, advanced construction and engineering techniques or increased fuel load following years of drought and fire suppression strategies.
‘Public model’ preferred
“Insurance companies want to use secret computer models so they can raise rates.” said Carmen Balber, executive director for Consumer Watchdog, a nonprofit consumer advocate organization.
“California needs a public model, open to public scrutiny and subject to public control as required by voters to guide the state in predicting the risk of wildfires caused by climate change,” Balber said in a written statement. “Private black box climate models are prohibited by Prop. 103 … because their findings and assumptions cannot be verified … defeating accountability. They do not address the most important issue – preventing death and destruction from climate change.”
Allowing insurance companies to use black box algorithms to make underwriting decisions and set insurance rates will legalize discrimination and immediately cause insurance rates to rise, according to Balber.
Doug Heller, director of insurance for the Consumer Federation of America, said he is confident a robust, transparent public risk assessment tool can be developed.
The American Property Casualty Insurance Association, The National Association of Mutual Insurance Companies, the Pacific Association of Domestic Insurance Companies and the Personal Insurance Federation of California recommend using an open, public risk assessment model that focuses on loss prevention that can best protect Californians against insurance company profiteering.
Seren Taylor, vice president of the Personal Insurance Federation, said there’s a need to update insurance filing regulations.
“Existing forward-looking risk assessment models in California are just used to assess earthquake potential in known fault zones and fires related to earthquakes. This should not be the only way to assess risk,” Taylor said.
“It is so bizarre that California is the only state that governs its insurance risks in this way. An exception to this is how wildfire risks are determined. Assessors look for locations based on where homes are in proximity to previous disasters.”
UC Cooperative Extension wildfire specialist Max Mority said catastrophe modeling is already being used in wildfire detection, evaluation, prevention and spatial modeling of land and structural features in aiding statistical analysis efforts and in the preparation of maps predicting wildfire zones.
“This approach involves a complex mix that includes data input credibility, model specifications, qualified uncertainties, whether findings are in line with assumptions, current science and limitations, and whether results are replicable,” Mority said.
Those are factors that can be applied in a public model, Mority said.
Data disclosure versus intellectual property protection
Roger Greiner is a senior vice president with Verisk Analytics, a provider of data-driven analytic insights to the insurance industry.
He noted that catastrophe models are critical to insurance risk analysis, as demand is growing to provide regulators with detailed information while keeping propriety information private.
Amy Bach, executive director of United Policy Holders, believes greater emphasis should be placed on offering insurance discounts rewarding fire risk reduction mitigation and prevention efforts when property owners clear brush and take other fire prevention measures.
However, the use of catastrophe models “could undo Prop. 103 protections,” according to Bach, who urged the state to be careful “since insurance companies support this model, they may expect to use it to increase rates.”
Bryan Rehor, with Zesty AI, said, “The role of an artificial intelligence model is to attempt to predict and/or mitigate natural disasters and assess their implications for society.” He said transparency is essential to show how assumptions are made and to validate findings.
Accelerating rate changes
Personal Insurance Federation President Rex Frazier characterized the state’s insurance regulation process as ”a slow-motion train wreck” referring to the time it takes for insurers to obtain rate change approvals. It can take 12 months to two years, as was the case with a rate request by AllState in 2021.
“The immediate priority is for DOI to work through all the rate filings pending in the system. The longer it takes for filings to get through, the longer it will take the market to heal.”
On the legislative front, he sees the immediate priority as recognizing the decay of the FAIR Plan, the state’s insurer of last resort, which is now the third-largest source of insurance only behind State Farm and Farmers.
“This is unsustainable. When the FAIR Plan runs out of money following a massive fire event, not only do the insurers need to pay customer claims, they need to add money to the FAIR Plan. This could impair small insurers and reduce large insurers’ capital, forcing them to cut back policies in force. A big FAIR Plan loss generating a deficiency assessment on insurers will trigger a mass nonrenewal of policies,” he said.