Decumulation Edge Cases — Annuities, Insurance, NQDC, Pensions

The retirement-income content we already publish covers the dominant decumulation problems: Monte Carlo for retirement, drawdown sequencing, Roth conversion windows, Social Security claiming optimization, RMD engineering after SECURE 2.0. Those are the well-trodden parts of the decumulation surface — and they're not where the bugs live.

The bugs live in the parts that aren't a single 401(k) drawing down through retirement. They live in the annuity riders that retirement-platform mock data treats as undifferentiated cash flows. They live in the HSA investment account whose qualified-medical-expense reimbursement timing breaks every projection engine that assumes immediate withdrawal. They live in the non-qualified deferred compensation arrangement whose §409A distribution elections fragment the planning horizon by a decade. They live in the defined-benefit pension lump-sum-vs-annuity decision that depends on actuarial-mortality assumptions most engines don't handle correctly.

This theme is the umbrella for the decumulation edge cases that the existing retirement-income content doesn't cover. Each piece below is a deep dive on one product or scenario most platforms quietly underhandle.

Why these products are the hard cases

Each of the four products covered in this theme has a feature that breaks the assumptions a domestic retirement-platform engine typically makes:

• Annuities introduce contract-level features (riders, surrender schedules, withdrawal corridors, mortality assumptions) that don't fit the position-and-cash-flow model that works for brokerage and 401(k) accounts.
• HSAs have the qualified-medical-expense-reimbursement deferral feature: an HSA holder can pay medical expenses out-of-pocket and reimburse from the HSA decades later, with the intervening period producing tax-free growth and no current taxable event. Most projection engines assume reimbursement happens contemporaneously with the expense.
• NQDC plans have §409A distribution-election rules that lock the holder into a specific distribution schedule years in advance, with severe penalties for changes. The plan assets are technically the employer's (subject to creditor risk) and don't appear on a personal balance sheet the way 401(k) assets do.
• DB pensions are an actuarial product, not an account-balance product. The "balance" is an accrued benefit translated to a present-value lump-sum offer at retirement; the value is sensitive to interest-rate assumptions, mortality assumptions, and the plan's funding status.

The four pieces under this theme

Annuity modeling

Annuity modeling: fixed, variable, indexed, SPIA walks through the four major annuity types, the data fields each requires (cash value, account value, surrender value, guaranteed minimum, riders), the rider taxonomy (GLWB, GMWB, GMIB, return-of-premium), and the §72(q) early-withdrawal interaction with the §72(t) retirement-account rules. Synthetic data has to model contract-level features, not just an aggregate "annuity value."

HSA investment modeling

HSA investment & triple-tax-advantage modeling is the deep dive on the HSA's distinguishing feature: the triple-tax advantage (deductible contributions, tax-free growth, tax-free qualified-expense distributions) and the qualified-expense-reimbursement deferral that turns the HSA into a stealth retirement account for tax-aware households. Test data has to model both the spending-account tier (most HSAs) and the investment-account tier (~10% of HSAs by count, much higher fraction of dollars).

NQDC / §409A

NQDC and §409A deferred-comp modeling covers the data shape of non-qualified deferred-compensation arrangements: deferral elections, distribution elections, the §409A constraints that lock the schedule, the employer-creditor-risk that makes NQDC qualitatively different from a personal account, and the SERP / supplemental-pension overlay common at senior-executive levels.

DB pension modeling

Defined-benefit pension modeling is the actuarial-product piece. Accrued benefit, normal retirement age, early-retirement reductions, joint-and-survivor options, lump-sum-vs-annuity offers, and the interest-rate sensitivity that makes lump-sum offers swing by 20-30% across a typical Fed cycle. Test data has to model the actuarial layer.

The lump-sum-vs-annuity comparison

Lump sum vs. annuity for pension decisions is the procurement-side comparison: the mortality-assumption framework, the discount-rate sensitivity, the inflation-protection question, the spousal-survivor consideration. The decision is one of the highest-stakes single decisions a retiree makes; the platform that supports it has to model both options correctly.

Supporting glossary terms

• SPIA (Single Premium Immediate Annuity) — the cleanest annuity case: lump-sum premium, immediate income stream, no cash value, no riders.
• Variable annuity — cash value linked to subaccount investment performance with rider-based guarantees.
• Indexed annuity — cash value linked to a market index (S&P 500, etc.) with caps, participation rates, and floors.
• HSA (Health Savings Account) — the triple-tax-advantaged account most retirement engines underuse.
• NQDC (Non-Qualified Deferred Compensation) — employer-promise-to-pay arrangements with §409A constraints.
• Defined benefit plan — the pension-style retirement plan whose benefit is a function of years of service and final salary.

Where this connects

The decumulation-edge-cases content interlocks with several other content threads:

• Retirement Income Sequencing — the existing theme covering the dominant decumulation problems.
• Time-Series Fidelity — annuity riders and DB-pension valuation depend on the same regime-switching return assumptions covered there.
• Cross-Border Wealth — annuities have specific cross-border tax treatment; pensions abroad add another layer.
• Equity Compensation — NQDC and equity comp share executive-compensation context but have different §409A vs. §83 mechanics.