Technology has been rushing headlong toward the future, no doubt. But if insurance carriers, agencies, MGAs, and MGUs have been sometimes slow to adapt, regulatory agencies are often slower, often making rules or laws after technological changes have overtaken the market.
The Connecticut Department of Insurance (DOI) under Commissioner Andrew Mais has taken the position that it doesn’t have to be this way. Instead, Connecticut’s DOI has restructured its data science and actuarial divisions to be more proactive with regulating the insurance market in the state.
Data science in insurance is nothing new
By reorganizing the department under Chief Actuary Wanchin Chou, the state hopes to keep better in step with the insurance industry’s progress as it adopts new data models. Chou began his career as a nuclear engineer before becoming an actuarial scientist, and his 30 years in the private insurance industry and regulatory paths have given him a very informed perspective on the shifting tide of information.
“This is nothing new. In 1997, the profession introduced credit scoring. In early 2000, big companies turned to predictive modeling and have used this for private passenger auto, homeowners, and commercial lines for decades now,” said Chou. “With social media, big data, AI, and a growing industry use of all of these combined with machine learning, we feel we need to restructure our resources in order to keep up with the industry and make regulation effective.”
Chou and Connecticut DOI Director of Communications Jim Carson said, despite Connecticut being “The Insurance Capital of the World,” the department runs on a fairly lean staff compared to the vastness of the industry presence in the state.
“We’re 150 highly skilled regulators doing world-class regulation,” Carson said.
That means the state DOI has taken a very forward approach, staying on the leading edge of regulatory trends while maintaining an experienced, if compact, staff who each have significant industry experience. Chou echoed the sentiment of Carson, saying that his staff of fewer than 10 actuaries and data scientists – while expected to grow somewhat – is “one of the best in the country.”
While operating on comparatively thin margins, Chou said the department doesn’t have the leeway to hire junior actuaries, so each staff member comes to the department with a weight of experience, often upward of 25 years.
“In the future, we are hiring more P&C actuaries and data scientists to leverage actuarial resources to not only meet the department mission but also address best practices in the industry,” said Chou. “Best practices should improve the overall effectiveness of regulating the industry and protecting consumers.”
How does Connecticut regulate insurance carrier data use?
Data regulation is one of those areas that has the potential for rife disparity state to state because not all state departments of insurance have rate-setting authority.
In Connecticut, the state has the authority to approve or deny rate increases and rate-setting standards among the insurance products that carriers offer within the state, in addition to more standard data requests for solvency assessments.
Along with this rate-setting authority comes the ability to request the data sources that compose an insurance carrier’s justification for pricing. Because insurers that want a rate change have to send the rate filing through the department, the actuaries of the department have the opportunity to analyze the actuarial standard of practice (ASOP) and challenge the consistency, the data, the variables, and sourcing that are foundational to an insurer’s model.
While the Connecticut DOI has historically kept actuaries on staff to handle solvency calculations and concerns, its interest in expanding the data science functions of the department stem from a sense of industry protection. It may be decades before we fully understand the impact of such vast data collection methods as artificial intelligence (AI), machine learning, internet-of-things (IOT) devices, social media, and other big data sources in the industry. Yet, the DOI isn’t interested in waiting those decades out to have a finger on the pulse.
“When we talk about the restructure of the department itself, of the actuarial and data science functions, it is because we want to work more collaboratively with the industry and have a more effective regulatory environment,” said Chou. “Data scientists need to have industry knowledge, and actuaries need to expand from their traditional functions of solvency and reserving calculations.”
This is, to Chou’s mind, the most effective approach to keep up with an industry that’s ripe for change.
“In the industry, more companies are using data scientists to pull statistics, trends, and other data from social media, and these sources have so much sensitive information,” said Chou. “In the insurance department itself, we need to pair data scientists with actuaries. Insurance companies are now using data scientists to build their models and pricing and more effective marketing, and it’s essential that we at the department also have data scientists who can peer-review these methods.”
That peer-review got an upgrade at the end of 2022, when the Connecticut DOI announced that, on an annual basis, insurance carriers in the state would need to submit their data practices to the state for “data certification.” In this data certification assessment, the state’s actuarial division will be able to analyze data sources, safety practices, and usage for its adherence to the state’s compliance standards.
Connecticut DOI Actuarial Division protecting both consumers and insurers
As Chou points out, adopting new methods to price business, maintain solvency, and capture market share for peak profits is nothing new in insurance. Yet, historical industry use of data tables and actuarial models has also been a source of very disparate outcomes that are implicitly or explicitly based on a consumer’s race or gender. That, Chou said, is one of the fundamental concerns underlying the department’s reorganization, that new innovations may continue to exploit the same historically disadvantaged groups.
In that way, Chou sees his department as aligning for both the interests of consumers as well as the interests of the insurance institutions themselves.
“Some of the unintentional bias in insurance and modeling is based on sensitive data. Collecting this data can subject insurers to criticisms that their aims are intentionally racist,” said Chou. “As a regulator, we can protect insurance companies from making this often-unintentional mistake by ensuring they don’t base products on this level of sensitive data, which in turn helps protect consumers.”
The Connecticut DOI is hoping that the state’s insurers see them not as an adversary but as a valuable partner. Requirements like the data certification and rate-setting review processes aren’t meant to be a venue for knuckle-wrapping correction, Chou said, but rather an opportunity for regulators and industry partners to find common ground on emerging technology and how to use it responsibly.
“I can’t say for other states, but generally I see departments of insurance need to open more communication and see the challenges from inside the industry’s perspective, to work with them, and to make sure these regulations are effective and not just troublesome,” said Chou. “We want to collaborate to provide a greater overall benefit to the industry. Insurance industry members and regulators have an interest to protect each other and consumers, and I think we set a good example of that in this state.”