The New York Department of Financial Services (“DFS”) recently released guidance describing how a life insurer may use so-called “external data,” that is, data not directly related to an applicant’s medical condition, without violating New York’s statutory prohibitions against unfair discrimination. Insurance Circular Letter No. 1 (“Circular”), DFS (Jan. 18, 2019).[1] The Circular also addresses transparency concerns attendant with such data. As the first directive of its kind, the Circular (i) places the burden of demonstrating non-discrimination on insurers, even when relying on third-party reports; and (ii) lays ground rules for life insurers considering the use of nontraditional data. Review of this Circular is vital for insurtech as well as established carriers that use or are considering use of nontraditional underwriting criteria.
Every state insurance department broadly prohibits unfair discrimination. Insurers know not to discriminate based on protected categories such as race, ethnicity, creed, or sexual orientation. Conversely, it is not controversial to “discriminate” based on the traditional set of criteria, e.g., age, height-weight ratio, medical examination results, family medical history, alcohol consumption, smoking, and driving record. Many types of data that do not directly implicate protected classes or traditional underwriting categories may be helpful in underwriting, but carriers have been reluctant to use such factors for fear they may act as proxies for prohibited criteria, and thereby run afoul of antidiscrimination regulations.
The Circular sheds light on when nontraditional metrics may constitute “unfair discrimination” by providing two basic prohibitions. First, a carrier may not use any data sources, algorithms, or predictive models that collect or utilize data on prohibited criteria, such as “race, color, creed, national origin, status as a victim of domestic violence, past lawful travel, or sexual orientation . . . or any other protected class.” DFS imposes this burden on insurance companies, even where that data derives from a third party or is proprietary in nature.
Second, such nontraditional data sources may not be used in a way that is unfairly discriminatory, as determined (at least in part) by a two-prong examination. The data or algorithm must be supported by (i) generally accepted actuarial principles or actual or reasonably anticipated experience justifying differentiation; and (ii) a valid rationale for differentiating rates accordingly. Thus, it is not enough that a facially-nondiscriminatory factor correlates with better or worse mortality. There must also be a clear rationale for that criteria or algorithm’s link to different mortality outcomes.
The Circular clarifies how New York’s unfair discrimination and consumer disclosure regulations apply to “external” or non-traditional data sources, algorithms, and predictive models. This guidance comes from a DFS investigation into life insurance underwriting practices that make use of nontraditional underwriting criteria. Although it contains cautionary tones about DFS’s attention to compliance with antidiscrimination and disclosure regulations, it also illuminates the path for carriers and their insurtech partners to incorporate nontraditional data sources that comply with DFS’s two-prong examination. We expect regulators from other states to take a similar approach to the use of nontraditional data in underwriting, and will stay attuned to any further developments in this emerging area of regulation.
[1] The Circular, “Use of External Consumer Data and Information Sources in Underwriting Life Insurance,” is available online at https://dfs.ny.gov/insurance/circltr/2019/cl2019_01.htm.
