NCOIL Genetic Testing Model Act: What Life Underwriting Faces in Spring 2026

The NCOIL Life Insurance and Financial Planning Committee closed public comment on its Life Insurance Genetic Testing Model Act on February 4, 2026 and is on track for a substantive vote at the Spring National Meeting.

The draft, sponsored by Rep. Brenda Carter of Michigan and advanced through two committee readings since June 2025, is the first serious model for the carve-out federal GINA left in place in 2008 for life, long-term care and disability income.

Key Takeaways

• Family history stays usable under Section 4(c), the single largest concession in the comment cycle and the reason accelerated underwriting survives the model in a way it does not survive Florida's 2020 statute.
• The voluntary disclosure carve-out is a one-way ratchet. A negative test can move an applicant to a better class; a positive one cannot surcharge or decline.
• 3 to 7 basis points of expected mortality load on a $500 million face block of ages 35 to 50 preferred nonsmoker female lives, at voluntary disclosure rates of 15 to 25 percent.
• 2 to 7 percent of pricing margin. Preferred-plus 20-year term runs 100 to 200 basis points of expected margin after acquisition, maintenance and reinsurance, so the load lands as a share of that, not of premium.
• Under 0.5 percent of the UK in-force book has had genetic testing as a material underwriting input across two decades of the ABI moratorium, the closest thing to a natural experiment the vote has.

Where the Draft Draws the Line

Section 3 defines a genetic test as an analysis of human DNA, RNA, chromosomes, proteins or metabolites detecting genotypes, mutations or chromosomal changes associated with a specific inherited condition. That deliberately excludes three inputs carriers ingest routinely: standard blood chemistry panels, prescription drug history, and diagnostic results evaluating a currently symptomatic condition.

The exclusion does real work. An A1c, an LDL reading, a PSA level or a liver function test is not a genetic test under this language, even though each has heritable components. A BRCA1 or BRCA2 panel run on an asymptomatic applicant is. A confirmatory BRCA test on an already-identified tumor is diagnostic, not predictive, and sits outside the definition. The draft keeps the predictive-versus-diagnostic line where underwriters have worked from it for fifteen years.

Section 4 then carves out information an applicant volunteers, but only in one direction: the carrier may use a negative result to improve a family-history-flagged applicant's class and may not use a positive result to surcharge or decline. That asymmetry is the whole actuarial question, and it is the same structure the UK ABI code has run on since 2001.

Against the seven state statutes already on the books, the model sits mid-distribution on substance and pulls harder on procedure.

The substantive gap with Florida is family history. Florida's language has been read to sweep it in, which forced carriers to pull family history questions or build a state-specific application workflow. Section 4(c) preserves it explicitly.

Sizing the Ratchet

Four variables set the load, and each is observable rather than assumed.

Population prevalence comes first: BRCA1 runs roughly 1 in 400 in the US general population and closer to 1 in 40 in Ashkenazi Jewish populations, with Lynch syndrome variants between that and the much rarer Huntington's allele. Consumer testing uptake is second: direct-to-consumer providers have reached roughly 30 million cumulative US customers, with hereditary cancer panel uptake concentrated in the ages 30 to 55 prime buying cohort. Third is the lifetime mortality differential a positive result implies over the decades most term policies underwrite. Fourth is propensity-to-select, the increment in application behavior driven by knowing the result, which practitioner literature puts at 20 to 40 percent.

Run those through a deterministic expected-value frame and a mid-size term writer booking $500 million of face on ages 35 to 50 preferred nonsmoker female lives should expect 3 to 7 basis points of additional expected mortality, at voluntary disclosure rates of 15 to 25 percent.

The number only means something against the margin it eats. Preferred-plus 20-year term typically carries 100 to 200 basis points of expected margin after acquisition, maintenance and reinsurance cost, so 3 to 7 basis points absorbs 2 to 7 percent of it. Carriers with exposure concentrated in BRCA-prevalent submarkets sit at the top of that range and rural-skewed books at the bottom, which is a dispersion that matters for block transfer pricing in 2026 and 2027.

The UK experience bounds it from the other side. Across two decades of the ABI moratorium, the share of the in-force book where genetic testing was a material underwriting input stayed below 0.5 percent, and premium impact across the protected bands is plausibly 1 to 3 percent on the preferred base, well under the double-digit predictions that preceded the 2001 agreement. Two adjustments carry it to the US: direct-to-consumer testing penetrated here five to seven years faster, compressing the absorption runway, while the NCOIL draft protects family history more explicitly than the ABI code does. Netting them puts full-phase-in impact at 1 to 4 percent on the preferred base.

The Cost Sits in the Pipeline, Not the Price

A 1 to 4 percent pricing load is absorbable on a term filing cycle that refreshes every 18 to 36 months. What is not absorbable on that schedule is everything the draft asks the underwriting stack to do.

The prohibition reaches downstream of the test result. A PARP inhibitor prescribed on a confirmed BRCA diagnosis is a pharmacy record, not a genetic test, but it is a genetic result wearing a different label, so Rx pipeline vendors have to build state-specific suppression logic by clinical use class.

The adverse action requirement compounds it. Producing a plain-language notice naming the factor behind a decline means pre-computing which score inputs are genetic-adjacent and which are not, before the model returns a decision. That is the architecture the Colorado AI Act consumer explanation regime already demands by June 30, 2026, and carriers that have not started it face a two-quarter build.

Reserving has a decision to make on a shorter clock than pricing does. VM-20 Section 9.C requires prudent estimate mortality reflecting anticipated shifts, and an adopted model law in a carrier's footprint is an anticipated shift. The year-end 2026 valuation therefore needs a documented choice: load for the projected selection effect now, or wait for observed experience. Carrying the legacy assumption forward without addressing the question is the one option that does not survive review.

Illustration work has the least runway of all. Term can wait for the next scheduled rate action, but universal life and whole life self-support testing under ASOP 24 runs quarterly and drives both the illustrated ledger and in-force cost-of-insurance discipline, so that assumption has to move as soon as a state adopts.

Reinsurers will reach the same conclusion independently. A ceding company without its own defensible mortality projection negotiates the next YRT renewal against the reinsurer's aggregate view of the regime rather than its own book. For new-business cohorts the assumption is a disclosure item too, as the LDTI first-full-year implementation notes set out: locked-in historic cohorts do not recalibrate, but everything issued after adoption is priced and reserved against the regime in force at the state of issue.