With spring’s arrival, a bouquet of differing NAIC groups and states is popping up to consider the use of big data and algorithms by insurers, including algorithms based on machine learning. Many are focusing on life insurance underwriting and seeking to ensure that any unfair bias is rooted out of the data and algorithms used by insurers. The NAIC’s 2022 activity will be cultivated by the newly formed Innovation, Cybersecurity, and Technology (H) Committee (H Committee). States are also blooming with activity, and Colorado planted its bulbs early in 2022 by holding two stakeholder meetings as required by Senate Bill 21-169 (as codified in Colorado Statutes section 10-3-1104.9).
NAIC
The new H Committee is facilitating all the NAIC groups and addressing innovation and technology issues so that information and insights can be cross-pollinated among all regulators and interested persons. The goals of the H Committee are to:
- Identify issues with the use of innovation and technology;
- Understand how the use of innovation and technology is affecting the insurance market;
- Understand how insurers are innovating and using technology; and
- Understand how such insurers’ use of innovation and technology can be regulated.
To further pollinate collaboration and to ensure no unfair bias takes root, one of the H Committee’s first projects is a collaboration forum that will (i) address algorithmic biases by identifying and addressing foundational issues and (ii) develop a common framework that can inform the specific workstreams in each NAIC group. This will bring together the work of the:
- NAIC Accelerated Underwriting Working Group (AU WG)
- NAIC Big Data and Artificial Intelligence Working Group (Big Data & AI WG)
These WGs also reported their activities at the NAIC Spring 2022 National Meeting.
The AU WG’s educational report fully bloomed, as it was adopted by the AU WG during the Spring National Meeting. The educational report provides a broad overview of life insurers’ use of big data and accelerated underwriting, grading the ground for regulators and interested parties. The educational report reviews the differences between accelerated underwriting, traditional underwriting, and simplified underwriting, as well as the current prevalence of these practices and expected trends for the future. It also reviews the use of various types of consumer data, including traditional data, non-traditional data, Fair Credit Reporting Act data, and the issue of using biased data.
Some consumer representatives criticized the educational report’s lack of concrete guidance for states and reliance on current unfair trade practices laws. However, the AU WG chair noted that the next work product of the AU WG would be to create a regulator guide that builds on the educational report and provides specific guidance for regulators.
Within the Big Data & AI WG, several workstreams are sprouting.
- Workstream One – Initially, the Big Data & AI WG sought to grow regulatory understanding of the use of artificial intelligence and machine learning in private passenger auto insurance; now, the workstream is branching out to conduct similar surveys for homeowners and life insurance.
- Workstream Two – Is seeking to grow tools to assist regulator review of accelerated underwriting models and help regulators determine whether bias is “baked into” the data or models being used.
- Workstream Three – Is studying the industry’s reliance on third-party vendors of data and algorithms and how to “best regulate these entities,” including through revised examination standards.
- Workstream Four – Seeks to germinate a white paper on a regulatory framework that brings together all the informational clippings from the other workstreams.
States
Oklahoma sprouted House Bill 3186 and Rhode Island sprouted House Bill 7230, which are substantially similar to Colorado Senate Bill 21-169. As we previously reported, Colorado prohibits insurers from using external consumer data, information sources, algorithms, or predictive models based on such data, in a way that unfairly discriminates based on race, color, national or ethnic origin, religion, sex, sexual orientation, disability, gender identity, or gender expression. Colorado has held two stakeholder meetings focusing on life insurance underwriting practices at which the key terms “external consumer data and information sources” and “traditional underwriting practices” and the required testing process were discussed.
Other state activity includes:
- New Jersey Assembly Bill 5651 requires annual reporting by automobile insurers using an automated or predictive underwriting system, to demonstrate that there is no discriminatory outcome in the pricing of insurance, and directs the commissioner of banking and insurance to cultivate rules and regulations.
- A preproposal statement of inquiry was planted by Washington regarding possible rulemaking on insurance underwriting transparency to address its concern that “insurance consumers are not provided with full disclosure and complete transparency from insurers for adverse actions, rate changes, or the factors that insurers consider in determining premiums.” The proposal would require insurers “to provide notices to consumers for all factors evaluated in any associated insurer actions, which must include an itemized disclosure of all variables considered in underwriting, as well as the proportionality or weight at which those factors were evaluated.”
- Connecticut updated its department notice concerning the “usage of big data and avoidance of discriminatory practices” to remind insurers of their obligation to ensure that their use of big data complies with federal and state anti-discrimination laws, regardless of whether the seeds for their data and algorithms are sourced internally or through a third-party vendor. Insurers are also required to submit an annual data certification to the Connecticut Insurance Department. The notice also asserts the Connecticut Insurance Department’s authority to require that insurance carriers and third-party data vendors, model developers, and bureaus provide the department with access to data used to build models or algorithms included in all rates, forms, and underwriting filings.
As spring turns to summer, more varietals are sure to emerge as other regulators begin tending their gardens. We will continue to monitor the activity of the NAIC and the states regarding insurers’ use of big data and algorithms.
