VM-22 Aggregation and the New Annuity Pricing Floor

Non-variable annuity pricing cannot be separated from reserve mechanics when aggregation rules, deposit-type contract scope, and settlement-option treatment change statutory capital strain at issue. The immediate pricing question is not whether VM-22 is a valuation project. It is whether pricing actuaries can convert the evolving reserve basis into an explicit annuity pricing floor before sales teams quote a fixed annuity, payout annuity, GIC-like contract, or pension risk transfer case on margins that look adequate only outside the statutory model.

The timing is concrete. The NAIC Life Actuarial (A) Task Force posted a June 2026 agenda to consider adoption of APF 2025-18 on VM-22 deposit-type contracts and APF 2025-20 on VM-22 aggregation, and to consider comments on a Group Annuity Selection Memo. The same task force exposed APF 2025-18 and APF 2025-20 in May 2026 after pending VM-22 subgroup action. From tracking NAIC valuation updates, this is the point where pricing teams should stop treating VM-22 as a back-office reserve implementation and start treating it as a live constraint on credited rates, option budgets, spread targets, and quote discipline.

The market makes that discipline more important. LIMRA reported final U.S. retail annuity sales of $464.1 billion in 2025, with fixed indexed annuities reaching $127.9 billion. In the first quarter of 2026, annuity sales cleared $100 billion for the tenth consecutive quarter, while fixed-rate deferred annuities still represented roughly one-third of the market despite a 16% year-over-year sales decline. High volumes, narrow competitive spreads, and rate-sensitive distribution mean the reserve floor can become the binding floor before the traditional profit margin does.

Start With Scope Before Margin

The first pricing control is a VM-22 scope map. A pricing memo that begins with target spread, commission scale, or credited-rate competitiveness is starting too late. The product cell needs to be classified by reserve category and contract feature before the pricing actuary calculates distributable earnings.

The 2026 NAIC Valuation Manual describes VM-22 as the principle-based reserve framework for non-variable annuities. It also sets the transition pattern: companies can elect VM-22 during the first three years after the effective date, but applicable blocks move to VM-22 on a prospective basis after the transition period. The manual's VM-22 language identifies payout annuity reserving, accumulation reserving, and longevity reinsurance categories, with pension risk transfer annuities included in the payout category. It also requires stochastic reserve calculations over a broad range of generated projection scenarios unless a deterministic reserve approach applies after testing.

That scope map has practical pricing consequences. A deferred fixed annuity with a multi-year guaranteed rate, a fixed indexed annuity with an option budget, an immediate annuity settlement option, a PRT annuity certificate, and a funding-agreement-like institutional product can have different statutory treatment even when sales teams discuss them as spread products. APF 2025-18 matters because deposit-type contract treatment affects whether GICs, synthetic GICs, funding agreements, and stable value structures sit inside the VM-22 model or remain under a simpler formulaic basis. The Academy's March 2026 GIC funding letter noted reasons to include those contracts, including consistency where companies manage them with products already inside VM-22, but also reasons for caution where products have limited policyholder optionality or existing asset adequacy frameworks.

For pricing, the answer is not to wait for every drafting item to settle. Build a pricing status field for each product cell: clearly in VM-22, potentially in VM-22, excluded but monitored, or out of scope. That field should feed the rate sheet. If a product is potentially in scope, the pricing floor should be run under both the legacy reserve basis and the VM-22 basis, with the higher reserve strain charge controlling until governance approves a lower treatment.

Aggregation Is a Pricing Input, Not a Valuation Footnote

APF 2025-20 raises the most direct pricing issue because aggregation can change reserve strain without changing the customer-facing economics of the contract. The Academy's aggregation letter identifies two meanings of aggregation: combining contracts and assets before scenario reserve calculation in a single model segment, or aggregating after scenario reserves are calculated and then applying the required tail measure. The Academy supported actuarially justified aggregation that reflects business management and economics, while warning that pre-calculation aggregation can create offsets that are not available in practice if blocks are managed separately.

That warning belongs inside the pricing model. Suppose a new accumulation annuity block creates liquidity exposure under excess surrender scenarios, while an older payout annuity block has shorter, more liquid assets. If the model nets the cash flows before the stochastic reserve calculation, the payout block may appear to fund liquidity needs from the accumulation block. If actual portfolio management, internal investment mandates, or regulatory constraints prevent those assets from being used that way, the aggregation benefit is not a pricing asset. It is a model artifact.

The useful pricing test is therefore two-sided. First, run the standalone contract cell or pricing cohort by issue year, guarantee period, surrender charge schedule, benefit form, and asset strategy. Second, run the same cell inside the permissible VM-22 aggregation structure. The difference is the apparent diversification credit. That credit should never disappear into a higher sales cap, a richer guaranteed rate, or a thinner quoted spread without an allocation rule approved by product, valuation, risk, and finance.

The 2026 Valuation Manual's VM-31 reporting requirements reinforce this discipline. They require disclosure of the impact of aggregation, including comparison of seriatim calculations to aggregation permitted under VM-21 or VM-22, and documentation of the methodology used to allocate aggregation benefit across model segments when aggregation crosses multiple segments. A pricing team can use that requirement as an operating standard: if the valuation report must document the aggregation benefit, the pricing file should document how much of that benefit is granted to the product, how much is retained as enterprise margin, and how much is unavailable for pricing because it depends on assumptions not controlled by the product line.

The Reserve-Aware Pricing Floor

A practical annuity pricing floor starts with present-value distributable earnings, not statutory earnings alone. The minimum spread or fee margin should be calculated as:

Minimum spread or fee margin = expected benefit cost + option cost + expense cost + default cost + capital charge + reserve strain charge - investment income credit.

Each component should be measured on a present-value distributable earnings basis under the relevant VM-22 grouping. The expected benefit cost captures guaranteed crediting, annuitization benefits, death benefits, income elections, and expected policyholder behavior. Option cost captures index-crediting budgets, caps, participation rates, floors, buffers, guaranteed living benefits where applicable, and hedge slippage. Expense cost includes acquisition expense amortization, trail commissions, maintenance expense, premium taxes, and any PRT implementation cost. Default cost reflects expected credit losses and downgrade costs on the asset portfolio. Capital charge reflects the target after-tax cost of required surplus. Reserve strain charge is the present value of statutory capital trapped by VM-22 reserves at issue and over the projection horizon. Investment income credit reflects the net asset earned rate available to fund those costs.

The pricing floor becomes binding when this sum exceeds the planned spread. In a fixed annuity, the response may be a lower credited rate, a shorter guarantee, a modified surrender schedule, or a lower commission allowance. In a fixed indexed annuity, the response may be a lower cap, participation rate, or option budget. In a payout annuity or PRT quote, the response may be a lower premium discount, stronger mortality improvement margin, higher asset default provision, or higher surplus load.

Patterns we have seen in recent annuity pricing briefs point to a common error: pricing teams measure capital strain after setting the competitive offer, then describe the result as acceptable if lifetime profits clear the hurdle. Under VM-22 aggregation, that sequence is backward. The pricing floor should be calculated first, before a rate or quote is released, because statutory strain can control distributable earnings in the early years even when ultimate economic value remains positive.

Standalone Cell Versus Aggregated Cohort

The right model output is not a single IRR. Pricing should show the standalone cell, the aggregated cohort, and the allocation of aggregation benefit. A compact template works:

Pricing view	Purpose	Decision use
Standalone seriatim cell	Measures reserve strain before diversification	Sets the conservative floor for new product features
Permissible VM-22 aggregation	Measures offsets by model segment and reserve category	Identifies credit available only if risk management supports it
Allocated pricing cohort	Applies an approved share of diversification benefit	Controls rate sheets, cap budgets, and PRT quote loads

The allocation rule should be mechanical enough to audit. One approach is to allocate aggregation credit in proportion to each cohort's standalone reserve strain contribution, capped by the amount of strain that cohort creates. Another is to retain a fixed enterprise share, for example 25% to 50%, and allocate the remainder by marginal contribution to conditional tail expectation. A third is to allocate no cross-category benefit to pricing unless the asset portfolios, ALM limits, liquidity plans, and risk monitoring are demonstrably integrated. The last approach is stricter, but it matches the Academy's concern that modeled offsets should reflect the way the business is actually managed.

Sensitivity Grids That Find the Binding Constraint

The pricing floor should be run as a grid, not as a point estimate. For accumulation products, the grid should vary guarantee duration, credited-rate cap, surrender charge schedule, utilization of free withdrawals, option budget, asset yield, reinvestment spread, and policyholder behavior. For payout and PRT business, the grid should vary mortality improvement, retiree age mix, form of payment, expense scale, asset default cost, reinvestment rate, and surplus commitment. For deposit-type or institutional contracts, it should vary credited-rate formula, withdrawal triggers, put options, spread reset mechanics, liquidity stress, and asset sale haircuts.

The output should identify the constraint that sets the floor: base reserve, stochastic reserve tail, additional standard projection amount, capital, liquidity, or distributable earnings. That distinction matters. If the base reserve controls, product changes that reduce optionality may not help much. If CTE tail scenarios control, reducing guarantees or changing surrender incentives may matter more. If capital is binding, the required spread may be driven by surplus cost rather than reserve level. If liquidity is binding, aggregation across a payout block may be unavailable even if it improves the model.

Credited-rate formulas deserve special attention under lower reinvestment spreads. A rate sheet may appear affordable at the current new-money yield, but fail the floor when renewal rates, option costs, or asset defaults move against the company. Fixed indexed annuity caps should be tested against higher implied volatility and lower bond yields at the same time, not independently. PRT quotes should test mortality improvement and asset default cost together, because aggressive longevity assumptions and private-credit-heavy asset strategies can both increase tail strain.

PRT and Group Annuity Quote Discipline

Group annuity and pension risk transfer quotes turn VM-22 from a product pricing issue into a selection issue. The June 2026 LATF agenda's Group Annuity Selection Memo item sits beside the VM-22 deposit-type and aggregation proposals for a reason: selection standards, reserve assumptions, and pricing assumptions interact.

For PRT, pricing actuaries should connect insurer-selection standards to quote assumptions for mortality improvement, retiree data quality, optional forms, expenses, asset default costs, and surplus commitment. A quote that relies on aggressive mortality improvement, narrow default spreads, and a thin capital charge may look competitive, but it can fail the reserve-aware floor if the VM-22 grouping produces early statutory strain. The pricing memo should show how the block will be modeled, whether aggregation with existing payout annuities is permitted, and whether asset portfolios are managed under a single ALM strategy or merely reported together for reserve purposes.

That last distinction is central. If a PRT block is priced on the assumption that payout annuity diversification offsets accumulation annuity surrender risk, the pricing file needs evidence that the insurer actually manages those risks through an integrated process. Without that evidence, the pricing floor should exclude the cross-category credit or retain it as enterprise margin rather than transferring it to the bid.

Governance: Keep Sales Moves Inside the Reserve Basis

The governance failure to avoid is simple: sales sees a competitor's higher cap rate or richer PRT bid, asks for a concession, and the pricing actuary updates the rate sheet without rerunning the VM-22 reserve basis embedded in the filed pricing model. Under a formulaic reserve regime, that may have been a manageable shortcut. Under VM-22, it can create a mismatch between pricing assumptions, valuation assumptions, and statutory strain.

A reserve-aware governance process should require four controls. First, every product cell should carry a VM-22 scope flag and aggregation basis. Second, every rate-sheet move above a materiality threshold should rerun the reserve-aware floor, including stochastic and capital output where relevant. Third, assumption changes for mortality improvement, lapses, withdrawals, default costs, expenses, hedge costs, and reinvestment spreads should be logged with owner approval. Fourth, any aggregation benefit used in pricing should be reconciled to the VM-31-style allocation documentation expected in valuation reporting.

This is not bureaucracy for its own sake. It prevents a hidden release of margin. If aggregation reduces reserve strain, the company can decide to pass some benefit to policyholders, retain some for enterprise capital, and use some to support product growth. But that decision should be explicit. A pricing floor that silently assumes full aggregation credit invites exactly the problem VM-22 documentation is trying to surface.

Why This Matters

For pricing actuaries, VM-22 aggregation changes the definition of competitive adequacy. The old question was whether the product clears a target margin under expected assumptions plus stress tests. The newer question is whether the product clears a reserve-aware floor under the statutory grouping that will actually be used to report reserves and support distributable earnings.

The practical answer is straightforward. Map scope first. Run standalone and aggregated reserve strain. Allocate diversification credit deliberately. Stress option budgets and credited-rate formulas under lower reinvestment spreads. Identify whether reserve, capital, liquidity, or earnings sets the floor. Then govern every sales-driven rate move against the same reserve basis. That is how VM-22 becomes a pricing control, rather than a compliance item discovered after the product has already been sold.