Net Unrealized Appreciation (NUA) — modeling the once-in-a-lifetime company-stock distribution decision

Net Unrealized Appreciation, codified at IRC §402(e)(4), is one of the highest-leverage tax provisions for retiring executives whose 401(k) holds employer stock. Done right, it converts what would otherwise be ordinary-income tax on the full distribution into ordinary-income tax on the cost basis and long-term capital-gain rates on the appreciation. The math can save high-NUA executives well into seven figures of federal tax.

Done wrong — and "wrong" includes "not exercised" — it forfeits that benefit permanently. NUA is a one-time election made at retirement. There is no do-over.

This is exactly the kind of decision retirement-planning platforms should be helping with. Most don't, because the data model required is more demanding than the typical 401(k) data feed surfaces. This article is the working note on what the model needs to look like, why the standard "company stock balance" field is insufficient, and what retirement-planning platforms ship without when the model is shallow.

What NUA actually is

When a participant in an employer-sponsored retirement plan (typically a 401(k) or ESOP) takes a lump-sum distribution that includes employer stock, they have a choice:

Option A — Roll over the entire distribution to an IRA. The stock comes out of the plan and into the IRA. No immediate tax. Future withdrawals from the IRA are taxed at ordinary-income rates regardless of how the stock subsequently performs.

Option B — Take the employer stock in-kind to a taxable brokerage account, applying NUA treatment. The participant pays ordinary-income tax in the distribution year on the cost basis of the stock — what the plan paid for it, often pennies on the dollar relative to current market value. The unrealized appreciation (NUA) is taxable only when the participant subsequently sells the stock, and at long-term-capital-gain rates regardless of holding period in the brokerage account.

The NUA election only applies if the distribution is a "lump sum" — within one tax year, the participant takes the entire account balance from all related plans of the same employer, after a triggering event (separation from service, age 59½, death, disability).

For an executive whose 401(k) holds $5M of company stock with a $200K cost basis, the math is dramatic. Option A taxes $5M at ordinary rates over the years they draw from the IRA — call it 32% federal + state, eventually around $1.8M+ in federal tax alone. Option B taxes $200K immediately at ordinary rates (~$64K) and then $4.8M of NUA at LTCG rates (~$1.14M assuming 23.8%). Net federal-tax savings: roughly $600K, plus the time-value benefit of deferring the LTCG tax until the participant chooses to sell.

The savings increase nonlinearly with the cost-basis-to-market-value ratio. A higher-NUA position produces more savings.

What goes wrong without the model

The standard 401(k) data feed shows participant-level account balance and asset allocation. Some feeds show employer-stock as a distinct holding. Few feeds show the cost basis of the employer stock — the single most important field for the NUA decision.

Without cost basis, the planning engine can't compute the NUA delta. It can't compare Option A to Option B. It can't surface the recommendation. It defaults to "we can't advise on this; consult your tax advisor for NUA decisions" — which is true but unhelpful for a decision that has to be made within one tax year of separation.

Beyond cost basis, several other fields are required:

• Acquisition history of the employer stock. Cost basis is per-share and varies across acquisition lots. Some shares were contributed via match in 2008 at a different price than shares purchased via ESPP in 2018.
• Pre-tax vs. after-tax stock distinction. After-tax basis in the plan (rare but possible) gets different treatment.
• Plan type. 401(k), ESOP, profit-sharing plan — same fundamental rule but different administrative paths.
• Related-employer plan coverage. The lump-sum requirement extends to all related plans of the same employer. A participant with a 401(k) plus a separate non-qualified deferred-compensation plan needs to know whether the NQDC plan affects the lump-sum determination.
• Triggering event status. NUA election requires a triggering event. The participant's status (separated, 59½, etc.) must be tracked.
• Prior IRA-rollover taint check. A partial rollover that's already happened can complicate the lump-sum requirement.

The decision tree

Once the data model supports the analysis, the platform can run the decision tree:

function evaluateNUA(participant: Participant): NUARecommendation {
  if (!participant.hasTriggeringEvent()) {
    return { 
      eligible: false, 
      reason: 'No triggering event', 
      futureCheck: nextLikelyTriggerDate(participant) 
    };
  }
  
  const employerStock = participant.plan.employerStockHoldings;
  if (!employerStock || employerStock.shares === 0) {
    return { eligible: false, reason: 'No employer stock' };
  }
  
  const optionA = projectIRARollover(participant, employerStock);
  const optionB = projectNUAElection(participant, employerStock);
  
  // Comparison must use after-tax NPV at expected drawdown rate
  if (optionB.afterTaxNPV > optionA.afterTaxNPV) {
    return { 
      eligible: true, 
      recommendation: 'Take NUA election', 
      delta: optionB.afterTaxNPV - optionA.afterTaxNPV,
      lumpSumWindow: '< end of distribution tax year',
      preconditions: lumpSumPreconditions(participant)
    };
  }
  
  return { 
    eligible: true, 
    recommendation: 'Roll over to IRA', 
    delta: optionA.afterTaxNPV - optionB.afterTaxNPV
  };
}

The comparison is sensitive to several inputs that vary by participant: expected drawdown timing (NUA savings increase if the participant plans to hold the stock longer), expected tax rate at drawdown (Option A's ordinary-rate hit is in the participant's drawdown bracket, which may be lower than current rate), expected investment return on the deferred LTCG taxes (if NUA is elected, the participant has the deferred-tax-on-NUA money invested), state of residence (NUA is a federal benefit; state tax treatment varies).

A lazy comparison that uses today's marginal rate for both options misstates the decision systematically — usually in favor of Option B, because the relevant rate for Option A is the future drawdown rate, often lower.

The lump-sum requirement and the partial-rollover trap

The NUA election requires that the entire account balance be distributed within one tax year. A common bug pattern: the participant or their advisor takes the cash portion of the 401(k) and rolls it to an IRA in December, then takes the employer stock in-kind in January with the NUA election. The two distributions span tax years; the lump-sum requirement fails; NUA election is invalidated; the entire stock distribution is treated as ordinary income.

Worse: the related-employer-plan rule. If the participant has any other plan of the same employer (ESOP, money-purchase, profit-sharing), that plan must also be distributed in the same tax year for the lump-sum requirement to be satisfied. Participants frequently forget about an old ESOP balance from a former subsidiary.

The platform must surface the lump-sum window prominently. Once the participant takes a single distribution from a related plan, the clock starts. The remaining distributions must complete in the same calendar year.

What the test corpus needs

A retirement-planning platform that wants to credibly handle NUA needs test data covering:

• Participants with low-basis employer stock (high NUA candidates)
• Participants with high-basis employer stock (Option-A candidates)
• Participants near triggering events (planning urgency)
• Participants with related-employer-plan coverage that complicates the lump-sum determination
• Participants who took partial distributions earlier in the year (lump-sum-trap case)
• Participants whose state taxation differs (FL no state tax vs. CA high state tax — affects the after-tax comparison)
• Participants whose expected drawdown rate differs (high-spending vs. low-spending retirees)
• Participants with concentrated employer-stock positions where NUA savings are dominated by concentration-risk considerations

Each scenario produces a structurally different right answer. A platform that exercises only "compute NUA delta on flat data" misses the cases that produce wrong recommendations.

The role of the synthetic corpus

Real retirement-data feeds rarely include cost basis with sufficient lot-level structure to power a NUA analysis. The platform may need to ask the participant directly, scrape it from the plan's records on file, or estimate from acquisition-pattern heuristics. A well-structured synthetic corpus contains the lot-level acquisition history, employer-stock cost basis, related-plan structure, and triggering-event status that real feeds typically lack.

This is a case where the synthetic corpus isn't just for testing — it's a template for what the platform should be capturing from real participants. If the synthetic corpus contains a field, the production data model should also contain it. If the platform's UI doesn't ask for the field at onboarding, it ships without the ability to advise on the decision the field powers.

What good NUA advice looks like

A retirement-planning platform that has the data model surfaces:

• For participants approaching retirement: a side-by-side comparison of the two options with the after-tax-NPV delta and the sensitivity analysis on the key assumptions (drawdown rate, tax rates, market return).
• For participants who just had a triggering event: an urgent flag with the lump-sum-window date and the steps to execute.
• For participants who already took a partial distribution: a flag indicating the lump-sum window may already be compromised, with the required follow-through actions.
• For participants accumulating employer stock: a long-term planning angle on whether the eventual NUA payout justifies the concentration risk of holding the stock in the plan.

The last point is the strategic surface most platforms miss. NUA isn't a retirement-day decision — it's a multi-decade allocation strategy. The platforms that surface this earlier in the participant's career are differentiated.

NUA is the prototypical case for why retirement-planning platforms need richer data than typical feeds provide. The decision has high financial stakes, narrow timing windows, and an analysis that the participant can't realistically perform alone. A platform that handles NUA well differentiates itself sharply from the field that defaults to "consult your tax advisor" for the highest-leverage retirement-day decision their participants will face.