The SOA Research Institute and LIMRA's 2020-2024 Individual Payout Annuity Mortality Experience Study, drawn from 26 companies representing over 80% of industry sales, 3.1 million contract-years exposed, and 143,190 recorded deaths, gives pricing actuaries their first look at annuitant mortality across the full COVID-19 arc (SOA Research Institute, 2026). The finding that matters for 2027 rate setting: a survivor-selection effect in the 2023-2024 tail suggests the annuitants who lived through 2020-2022 now run below the 2012 IAM Table's mortality projection, which raises SPIA net single premiums just as sales hit $3.7 billion in Q1 2026, up 22% year over year (LIMRA, May 2026).

A Study Built to Span the Whole Pandemic Arc

The prior individual payout annuity mortality study covered 2014-2019, a stable pre-pandemic baseline with no reason to expect a discontinuity in annual mortality experience. The new study is different by design: it opens in 2020, runs through the 2020-2022 excess-death shock, and closes in 2023-2024, the first two full years of what the industry has taken to calling the post-pandemic period. That span is what makes the dataset useful for pricing rather than merely descriptive. A study confined to 2020-2022 would show elevated mortality and nothing else. A study confined to 2023-2024 would show a rebound with no reference point. Only the combined five-year window lets a pricing actuary see the shape of the curve, and the shape, not the average, is the actionable signal.

The study benchmarks actual experience against three references: the 2012 Individual Annuity Mortality Table, the prior 2014-2019 study, and U.S. population mortality (SOA Research Institute, 2026). That triangulation matters because annuitant mortality and population mortality diverge structurally, a healthy-annuitant effect embedded in every payout annuity block since underwriting selection favors longer-lived purchasers. U.S. population mortality ran at roughly a 1.20 observed-to-expected ratio during 2020-2022, concentrated at ages 70 and older (CDC excess mortality data, as cited in Milliman's COVID-19 mortality trend analysis, 2026). Annuitants skew older than the general population by construction, so a naive read might assume their A/E ratio moved in lockstep. It likely did not, and the size of that gap, not its direction, is what the new study exists to quantify for this specific population rather than infer from population-level data.

Building the A/E Ratio, Cell by Cell

The mechanical core of any mortality experience study is the actual-to-expected ratio, computed separately for each age-gender-duration cell rather than as a single portfolio-wide number. Expected deaths in a cell come from applying the 2012 IAM Table's qx value, adjusted by the applicable projection scale, to the contract-years exposed in that cell. Actual deaths come from the study's 143,190 recorded deaths, allocated across the same cell structure. Dividing actual by expected produces the A/E ratio, and a ratio below 1.0 means annuitants in that cell are dying less often than the table predicts, which is an adverse outcome for a company that priced assuming the table.

Applied across 2020-2024, the cell-level A/E ratios should trace something close to a V-shape. During 2020-2022, COVID mortality pushed actual deaths above the table's expectation in the older age bands where the disease's case fatality rate concentrated, meaning the A/E ratio likely ran above 1.0, a temporarily favorable result for SPIA issuers holding those liabilities. By 2023-2024, the mechanism reverses. COVID disproportionately killed annuitants with pre-existing comorbidities, the population most likely to have been priced as standard or even substandard risk within the annuitant pool. The annuitants who survived that selection pressure are a health-screened subset of the original cohort, and a health-screened subset should, all else equal, exhibit lower mortality than the table assumes for their age and duration, pushing the A/E ratio in the 2023-2024 cells below 1.0.

The pricing actuary's job is not simply to read the ratio off the study but to decompose it. How much of a sub-1.0 A/E ratio in 2023-2024 is structural, a permanent shift in the surviving cohort's underlying health composition, versus transient, a short-term reversion as the population's mortality experience normalizes after two years of elevated volatility? The study's five-year span is precisely long enough to start separating those two effects, since a purely transient rebound should compress toward the pre-pandemic 2014-2019 baseline by the second post-shock year, while a structural survivor effect should persist or even widen.

Credibility Weighting for Companies Below Full Credibility

Few individual companies write enough payout annuity business to reach full statistical credibility in every age-gender-duration cell, which is exactly why an 80%-of-market industry study exists as a pricing input rather than a curiosity. The standard approach applies the classical square-root credibility formula, Z = min(1, √(n/nfull)), where n is the company's own observed deaths in a cell and nfull is the deaths required for full credibility, conventionally set around 1,082 for a 95% confidence interval with a 5% margin of error. A company with 270 observed deaths in a given cell earns Z = 0.5, meaning its filed assumption blends 50% of its own experience with 50% of the industry A/E ratio from the SOA-LIMRA study.

Smaller SPIA writers, and most writers are small relative to the handful of companies that dominate industry volume, lean more heavily on the industry study than on their own book in the cells that matter most for pricing: the older-age, longer-duration cells where a single company's exposure thins out fastest. That makes the 2020-2024 study's cell-level detail, not just its headline A/E figures, the operative input for any company below full credibility, which in practice is most of the market outside the top handful of payout annuity writers.

Choosing an Improvement Scale After a Five-Year Disruption

Mortality improvement scale selection was a comparatively quiet assumption before 2020: actuaries applied a long-run improvement vector, checked it periodically against emerging experience, and moved on. The pandemic broke that routine, and three distinct approaches are now live in the market. The first treats the SOA MIM 2026, the Mortality Improvement Model updated with four additional years of annuitant data, as the new base improvement scale without further adjustment, on the theory that MIM 2026 already embeds whatever the pandemic did to the trend. The second applies an explicit COVID load layered on top of a pre-pandemic improvement scale for projection years 2020-2024 specifically, isolating the disruption rather than blending it into the long-run trend. The third, a revert-to-trend approach, discounts the 2020-2024 volatility entirely once 2024 closes and resumes the pre-pandemic improvement vector for all future years, betting that the pandemic was a one-time shock with no lasting effect on the improvement rate itself.

Each choice embeds a different view of the survivor-selection question. MIM 2026's baked-in approach implicitly treats the observed 2023-2024 experience as informative about the going-forward trend, consistent with a structural read. The revert-to-trend approach implicitly treats it as noise, consistent with a transient read. There is no version of this decision that avoids taking a position on the same question the A/E decomposition above is trying to answer directly, which is why the improvement scale choice and the A/E interpretation should be argued together in a single assumption memo, not signed off separately by different reviewers.

From A/E Shift to Net Single Premium

The net single premium for a life-only SPIA paying $1 per year is äx = Σ vk · kpx, summed over each future year k, where v = 1/(1+i) is the discount factor and kpx is the probability that an annuitant aged x survives k years. A downward shift in the A/E ratio, annuitants living longer than the table implies, raises every kpx term in the summation, which raises äx, which raises the net single premium the insurer must charge to fund the same promised income stream. For a 70-year-old female priced at a 5% discount rate, a 1% improvement in survival probability translates to roughly a 0.5% to 0.8% increase in net single premium, a relationship that compounds across the summation rather than applying uniformly to a single year's cash flow.

Scaled up, a 2-percentage-point adverse shift in the A/E ratio, meaning annuitants live about 2% longer than the 2012 IAM Table with the current improvement scale implies, increases the net single premium of a $1,000-per-year SPIA for that same 70-year-old female by roughly $80 to $100 (based on standard annuity-due present value mechanics under a 5% valuation rate). That is not a rounding error on a product where insurers compete on quoted rate to the second decimal place. A carrier that has not updated its mortality basis to reflect the study's post-COVID survivor signal is either underpricing new business by that margin, compressing gross margin on every policy issued at the old rate, or is unknowingly holding an inadequate reserve on business already on the books if the pricing basis and the reserving basis share the same stale assumption.

The VM-22 Reserve Adequacy Question

Principle-based reserving for non-variable annuities under VM-22 requires a best-estimate mortality assumption, calibrated with margin, not a prescribed factor lookup (NAIC Valuation Manual, January 1, 2026 Edition). The 2020-2024 SOA-LIMRA study is now the most current, most defensible basis available for calibrating that best-estimate assumption for payout annuity business, which puts a real deadline behind an otherwise academic mortality-study release: VM-22 becomes mandatory for all new non-variable annuity issues by January 1, 2029, and companies still calibrating reserves off the 2014-2019 study, or off the 2012 IAM Table with no COVID-era adjustment at all, are working from an assumption set the industry's own experience data no longer supports.

The mechanical risk is specific. VM-22's deterministic reserve is generally set at a margin above the net single premium under a conservative mortality scenario, and that margin was calibrated against a pre-pandemic view of how much a best-estimate assumption could plausibly miss. If the structural survivor effect the new study surfaces lowers best-estimate mortality by 2% to 3% relative to what a company's current basis assumes, the prescribed margin no longer spans the gap between best-estimate and adverse-scenario mortality the way it was designed to. A company that treats the new study as a routine periodic update rather than an input that may require re-running its VM-22 reserve adequacy test is assuming the margin still covers a gap that the data suggests has widened.

Why This Matters

Every payout annuity mortality study since 2012 has been an incremental refinement of a stable trend. This one is not. It is the first industry benchmark built to capture a genuine discontinuity, an excess-mortality shock followed by a survivor-selection effect in the same five-year window, and it arrives while SPIA and DIA sales are both growing and while VM-22's mandatory 2029 deadline is close enough to be a current planning input rather than a future one. Pricing actuaries setting 2027 SPIA rates need a documented position on three linked questions: how much of the 2023-2024 A/E ratio is structural versus transient, which improvement scale assumption that position implies, and whether the current reserve basis's margin still covers the resulting best-estimate shift under VM-22. Treating the study as a routine data refresh rather than a signal that the mortality basis itself needs re-examination is the mistake this release is designed to prevent.

Further Reading

Sources

  1. SOA Research Institute, “2020-2024 Individual Payout Annuity Mortality Experience Study,” SOA.org, 2026
  2. LIMRA, “U.S. Annuity Sales Top $107 Billion in First Quarter 2026,” LIMRA.com, May 2026
  3. LIMRA, “Final U.S. Retail Annuity Sales Set New Sales High, Totaling $464.1 Billion in 2025,” LIMRA.com, 2026
  4. NAIC, Valuation Manual, January 1, 2026 Edition, VM-22 Non-Variable Annuity Reserve Requirements, content.naic.org, 2026
  5. Milliman, “Potential Effects of COVID-19 and Mortality Trends on Defined Benefit Plans,” Milliman.com, 2026
  6. Milliman, “Current State of Principle-Based Reserving for Non-Variable Annuities (VM-22),” Milliman.com, 2026
  7. SOA / LIMRA, Individual Payout Annuity Mortality Experience Study flyer, Experience Studies Pro, 2026