Circular Letter No. 1, issued by the New York Department of Financial Services (DFS) on January 18, 2019, sets out two "principles" that DFS expects life insurers to follow when they use external data sources, algorithms, or predictive models in underwriting or rating. The principles articulated in the Circular have profound implications for the use of external non-medical data sources and, thus, for the advancement of accelerated and automated underwriting that relies heavily on such sources and the use of artificial intelligence.
Citing concerns about the potential for unfair discrimination and lack of consumer transparency, the Circular imposes heightened diligence and disclosure standards on life insurers that use external non-medical data sources. The Circular prohibits life insurers from relying on a vendor's representations and requires that they "independently determine" the external data sources do not "collect or use" impermissible criteria, and are not unfairly discriminatory, notwithstanding the proprietary nature of the vendor's data or process. The Circular also requires complete transparency by the insurer to the consumer by providing material elements of the underwriting process and the external data sources upon which the insurer relies.
The issuance of Letter No. 1 follows an investigation initiated by DFS in June 2017 into the use of external consumer data or information sources not directly related to the applicant's medical condition to supplement traditional medical underwriting or as part of an accelerated or algorithmic underwriting program. The articulated purpose of the Circular "is to advise insurers authorized to write life insurance in New York of their statutory obligations regarding the use of external consumer data and information sources in underwriting for life insurance."
The Circular acknowledges the value of innovation and technology in the insurance sector, noting its potential to benefit insurers and consumers alike "by simplifying and expediting life insurance sales and underwriting processes" and that external data sources can lead to "more accurate underwriting and pricing of life insurance."
However, in a page taken almost verbatim from consumer advocates at the NAIC DFS expresses concern regarding wide variations in the accuracy of data sources, the absence of regulatory oversight and consumer protections, the potential for consideration and use of illegal factors, and the lack of transparency for consumers, without explaining the factual bases for those concerns. To that end, the Circular addresses both "Unlawful Discrimination" and "Consumer Disclosure/Transparency," outlining how DFS intends to apply NYIL Articles 26 and 42 to the use of external data sources.
Unlawful discrimination and disparate impact analysis
The Circular makes the broad pronouncement that the "N.Y. Insurance Law, Executive Law, General Business Law, and federal Civil Rights Act" generally govern the activities of insurers, including the ability of insurers to underwrite based on certain criteria." Relying specifically on Article 26 and §4224, the Circular identifies those characteristics that it states can never be used in underwriting (eg, race, creed, national origin) and those that cannot be used without actuarial validation.
The Circular signals that these will be the enforcement criteria that DFS will apply and the corresponding obligations it will impose on insurers using external data sources. That approach seems to follow the profound skepticism that the Circular expresses as to the validity of the use of non-medical data in life underwriting:
Based on its investigation, the Department has determined that insurers' use of external data sources in underwriting has the strong potential to mask the forms of discrimination prohibited by these laws. Many of these external data sources … have the potential to reflect disguised and illegal race-based underwriting that violates Articles 26 and 42.
Other models and algorithms … may either lack a sufficient rationale or actuarial basis and may also have a strong potential to have a disparate impact on the protected classes identified in New York and federal law.
DFS identifies the "principles" that it expects life insurers to follow before using such sources. Specifically, an insurer may not use an external data source, algorithm or predictive model in underwriting or rating, unless:
- The insurer "has determined that the external tools or data sources do not collect or utilize prohibited criteria" and
- The insurer "can establish" that the criterion being used is not unfairly discriminatory in that (a) its use is "supported by generally accepted actuarial principles or actual or reasonably anticipated experience; and (b) there is a "valid rationale or explanation supporting the differential treatment of otherwise like risks."
The Circular provides no details, however, as to how an insurer might show it meets these tests.
The Circular also takes a dangerous step away from risk-based underwriting and pricing with the conclusion that statistical and actuarial validation of a "guideline" that is based on external data sources or information" is not sufficient. It appears that DFS may subject an otherwise valid guideline to a disparate impact analysis and require proof of "a valid rationale or explanation supporting the differential treatment of otherwise like risks," again without explaining how an insurer is expected to comply, much less the statutory basis for any such compliance standard.
Consumer disclosure and transparency
The Circular provides that in those circumstances in which an insurer is required to provide an insured with a reason for an adverse underwriting decision, a life insurer that based its adverse decision in whole or in part on "external data sources or predictive model" must include "details about all information upon which the insurer" acted "including the specific source of the information upon which the insurer based its adverse underwriting decision." An insurer cannot "rely on the proprietary nature of a third-party vendor's algorithmic processes to justify the lack of specificity related to an adverse underwriting action."
More significantly, DFS takes the position that an insurer's obligation to provide such information under NYIL § 4224(a)(2), is not limited to declinations, limitations, and rate differentials, but also includes "other adverse underwriting decisions" the scope of which is broad enough to include "the inability of an applicant to utilize an expedited, accelerated or algorithmic underwriting process in lieu of a traditional medical underwriting."
The Circular also suggests that it may be an unfair trade practice to fail "to adequately disclose the material elements of an accelerated or algorithmic underwriting process, and the external data sources upon which it relies, to a consumer …." From a practical standpoint, DFS' point seems targeted at discouraging insurers from using non-traditional, non-invasive methods to extend the amount of insurance a consumer can acquire by validating unverified health data supplied by the consumer.
The Circular should be seen as a precursor to further action, including targeted enforcement investigations, by DFS regarding the use of non-medical external data sources in life underwriting. DFS has made it clear that it expects insurers to independently audit those data sources to assure they do not collect or use impermissible data and to verify not only the actuarial soundness of guidelines that use that data, but also evaluate whether those guidelines have an adverse disparate impact on protected classes and, if so, whether there are neutral rationales for the guideline.
The principles articulated by DFS in the Circular are premised on novel and, at best, debatable legal theories and will require thoughtful analysis by life insurers who are now constrained to examine their existing underwriting and rating programs in light of those principles.
Learn more about the implications of the Circular by contacting Peter Rice or Nicholas Kourides.