Deferred Compensation: What 409A Means for Employers

An advisor once told me the way to hold on to my best person was to promise them money later. Not equity, not a raise. A deferred compensation plan, structured properly, that would pay out in five years and give them a reason to stay for all five of them.

It sounded elegant. I spent about three weeks on it before working out that the version I had in my head was illegal, the version that was legal cost more in legal fees than the amount I wanted to defer, and the person I was trying to keep would have been holding an unsecured claim against a company that had eleven months of runway. I paid a cash retention bonus instead and it worked.

This is the article I wanted then: what nonqualified deferred compensation is, what section 409A does to it, who pays when it goes wrong, and an honest account of the narrow set of circumstances where a small business genuinely needs one. I build the people and records tooling for businesses without an HR department at FirstHR, which is an onboarding and HR platform rather than a payroll or retirement plan provider. This is general information and not tax or legal advice.

What It Actually Is

Nonqualified deferred compensation is a binding promise to pay an employee in a future tax year for work performed now, made under an arrangement that deliberately fails the requirements for a qualified retirement plan.

That last point is where small businesses meet these rules for the first time. Nobody sets out to build a deferred compensation plan. They write a sentence in an offer letter promising a payment in three years if certain things happen, and the sentence creates one.

The word nonqualified is not a criticism. It describes a plan that has not met the qualification requirements of the tax code, and therefore does not get the protections or the tax treatment that come with them. Everything that follows is a consequence of that single fact.

Qualified Plans vs Nonqualified Promises

A qualified plan holds the employee’s money in a trust that your creditors cannot touch. A nonqualified plan holds nothing at all, because the moment the money is genuinely set aside for the employee, they are taxed on it.

That is the entire distinction, and it drives every other difference. Once you accept that the assets have to stay yours and stay at risk, the rest of the design follows automatically.

The 2026 figures come from the annual cost of living adjustments in IRS Notice 2025-67. The absence of a limit in the right hand column is the whole appeal: somebody already deferring the full $24,500 into a 401(k) has no further room inside a qualified plan, and this is the structure that creates more.

The deduction row is the one that ends most conversations early, and it deserves more weight than it usually gets. A contribution to a qualified plan is deductible now. A nonqualified deferral is deductible only in the year the employee is taxed on it, which is the year it is paid. You accrue the expense on your books for a decade and get no tax relief against it until the cash finally leaves.

Before reaching for this structure it is worth confirming the qualified plan is actually full. Testing constraints stop many owners well short of the statutory limits, and the two standard answers to that are a safe harbor design or a hard look at why the plan is failing nondiscrimination testing in the first place. Both are cheaper than what follows.

Section 409A and Why It Exists

Section 409A is the set of operating rules that a nonqualified plan has to follow to keep its deferral, and it does not care how much you defer. It cares about when the decision was made and when the money is allowed to move.

The provision arrived after a series of corporate collapses in which senior people accelerated their deferred balances out of failing companies while everybody else waited in line. Congress responded by removing discretion from the timing of these arrangements altogether. The statutory framework sits at 26 U.S.C. 409A.

Understanding the intent helps, because the rules feel arbitrary until you see what they are guarding against. Every one of them exists to stop somebody choosing, after the fact, when their tax bill arrives.

Notice what is absent from that list. There is no limit on the amount, no coverage test, no requirement to offer anything to anybody else. The cost of a nonqualified plan is not economic. It is the cost of never getting the paperwork or the administration wrong, in either direction, for as long as the plan exists.

The Election Timing Rule

The deferral election must be made before the start of the tax year in which the compensation is earned, and there are only two exceptions to that.

The first is a newly eligible participant, who has 30 days from the date they first become eligible to elect, and only in respect of compensation for services performed after the election. The second is performance-based compensation earned over a period of at least twelve months, where the election can be made up to six months before the end of that period, provided the outcome is not substantially certain by then.

Read that back and you can see the practical shape of it. Deferring next year’s salary is a decision made this December. Deferring a bonus for a performance year running from January to December can wait until June 30 of that year, but not to the day the bonus is calculated.

The most common real failure I have seen described is not exotic. A company decides in the fourth quarter that a discretionary bonus should be paid over three years to encourage somebody to stay. The intention is fine. The timing is fatal, because the compensation was earned in a year for which the deferral election window closed the previous December.

Changing your mind later is worse. A subsequent election to push a payment further out has to be made at least twelve months before the scheduled payment date, cannot take effect for twelve months after it is made, and must delay the payment by at least five additional years. Three constraints, all of them cumulative, on what feels like an administrative change.

The Six Permitted Payment Events

A compliant plan may only pay on six events, and they must be specified in the document at the time of the deferral rather than chosen later.

Two of these rows deserve more attention than they usually get. Separation from service has its own regulatory definition built around the level of services still being provided, so keeping a departing executive on a consulting retainer can mean the payment trigger has not happened. And a change in control is a defined term, not a synonym for anything that feels like a liquidity event.

Then there is the anti-acceleration rule, which catches decent employers more often than cynical ones. Once a payment is scheduled, you cannot move it forward. Not for hardship that falls short of the emergency definition, not because the employee asks, not because the board agrees it would be fair.

The six month delay for specified employees is the rule that is most often applied where it is not required. It bites where a company has stock publicly traded on an established securities market or otherwise and a key employee separates, in which case payment waits half a year. A privately held small business generally sits outside it entirely, which is worth confirming with counsel rather than assuming in either direction.

What a Failure Costs

A failure triggers immediate income inclusion of everything vested under the plan, an additional tax of 20 percent, and premium interest running back to the year of deferral. Every dollar of it is charged to the employee, not to you.

Sit with that for a moment, because the asymmetry is the reason this section exists. The person who drafted the plan, chose the payment terms and administered the elections is the employer. The person who receives a tax bill for money they have not been paid is the employee.

Put numbers on it and the shape becomes obvious. Somebody with $300,000 of vested deferrals built up over six years discovers a defect in the payment provisions. The whole $300,000 is income this year at their marginal rate. The additional tax alone is $60,000. Premium interest is charged as though each year’s deferral had been taxable when made. The plan itself has paid out nothing.

Almost every employer in that position ends up making the participant whole, which is the quiet way the cost migrates back to the business. A gross-up on a penalty of that size is itself taxable compensation, so the true cost is materially higher than the penalty. Budgeting for a compliance risk that formally sits with somebody else is uncomfortable, and it is the realistic planning assumption.

There is a further sting in the aggregation rule. The regulations group arrangements of the same type, so one defective agreement can pull an employee’s other deferrals of the same category into the failure with it. A single badly worded severance clause can therefore contaminate a supplemental retirement balance built up over a decade.

The IRS does run correction programs. Notice 2008-113 covers operational failures and Notice 2010-6 covers document failures, and both can reduce or eliminate the consequences if the problem is found and fixed in time. Neither is available once an examination has started, and both are considerably more expensive than drafting the document correctly.

The Top Hat Exemption

A nonqualified plan escapes most of ERISA only if it is unfunded and maintained primarily to provide deferred compensation for a select group of management or highly compensated employees. That is the top hat exemption, and without it these plans would be impossible to run in practice.

What the exemption buys is significant. It removes the plan from the ERISA rules on participation and vesting, on minimum funding, and on fiduciary responsibility. A plan that failed the test would have to vest on a statutory schedule and fund itself like a pension, which is precisely the outcome the nonqualified structure exists to avoid.

Reporting and disclosure obligations survive, but they are light. A one-time statement filed with the Department of Labor within 120 days of the plan becoming subject to the rules satisfies them, and since 2019 that filing has to be made electronically through the agency’s system (Department of Labor). One form, once, and no annual return.

The difficulty is that nobody has ever defined a select group. There is no percentage, no compensation threshold, and no safe harbor. Courts have looked at the proportion of the workforce covered, at how participant pay compares with everybody else, and at whether the people involved had the bargaining power to influence their own terms.

In Demery v. Extebank Deferred Compensation Plan (2d Cir. 2000), the leading appellate case on the point, a plan open to 15.34 percent of the employer’s workforce was upheld, with the court noting that this figure was probably at or near the upper limit of the acceptable size for a select group. That is the closest thing to a number anybody has, and it is a ceiling rather than a target.

The practical risk for a growing business is drift. A plan drawn up for three executives quietly expands to cover every director, then every manager, and the exemption it depends on erodes without anybody making a decision. Write the participant list down, review it annually, and treat each addition as a legal question rather than a fairness one.

The Rabbi Trust Problem

A rabbi trust holds the assets earmarked for a nonqualified plan and protects them from you changing your mind. It cannot protect them from your creditors, and the moment it does, the deferral collapses.

The mechanism is a grantor trust drafted on model language the IRS published in Revenue Procedure 92-64. The employer contributes assets, the trustee holds them, and the plan document says what they are for. What makes it work is the clause that participants dislike most: in the event of the employer’s insolvency, the trust assets are subject to the claims of the employer’s general creditors, and the trustee must suspend payments once it has notice.

That last point deserves emphasis because it runs against instinct. A clause that funds the trust automatically if the company gets into difficulty looks like sensible protection for participants. It is a taxable event, by statute, at the moment the plan first provides for it.

Which brings the honest question forward. If the money cannot be protected from your creditors, what exactly is the employee holding? An unsecured claim, ranking with your trade suppliers, on a business whose survival over the deferral horizon is a matter of judgement. For a stable, profitable, decades old company that is a reasonable thing to accept. For a business that has raised money and is still finding its footing, it is not much of a benefit at all.

The FICA Special Timing Rule

Social Security and Medicare tax on deferred amounts is due at the later of the date the services are performed or the date the amount is no longer subject to a substantial risk of forfeiture, which usually means the vesting year rather than the payment year.

This is the one genuinely favorable rule in the whole structure, and it is set out at 26 CFR 31.3121(v)(2)-1. Take the amount into account when it vests and a non-duplication rule keeps that amount, and every dollar of investment growth on it afterwards, out of FICA wages permanently.

The arithmetic is better than it sounds. A senior employee is usually already above the Social Security wage base, $184,500 for 2026, by the time the deferral vests, so the only cost that year is Medicare tax on the deferred amount. Fifteen years of compounding then leaves the plan free of any further Social Security or Medicare charge.

Miss the vesting year and the default rule applies instead. The full balance, including all that growth, becomes FICA wages when it is paid, at the rates and wage base then in force. The difference on a mature balance is not small, and correcting it after the employment tax limitation period has run is generally impossible.

The part employers underestimate is who carries the loss. In Davidson v. Henkel Corporation (E.D. Mich. 2015), a federal court held an employer liable to retired participants for the reduced value of their benefits after it failed to apply the special timing rule, on the ground that the terms and purpose of its own plan required it. Getting payroll timing wrong on a nonqualified plan is not just the participant’s problem.

What a Bonus Plan Does Instead

Most of what a small business wants from deferred compensation is retention over two or three years, and a vesting cash bonus paid inside the short-term deferral window achieves that with no exposure to section 409A at all.

The rule is precise and generous. A payment made no later than the fifteenth day of the third month following the end of the year in which the amount vests is a short-term deferral and sits outside the regime entirely, under 26 CFR 1.409A-1. For a calendar year business, that date is March 15.

So a retention award that vests on December 31 and is paid the following February is a bonus. You can attach a three year cliff to it, forfeit it on early departure, split it into tranches with separate vesting dates, and never once touch these rules. What you cannot do is let the employee decide when to take it, because that choice is the thing that creates a deferral.

The severance line is worth a separate note, because severance is deferred compensation more often than employers realize. Involuntary separation pay escapes section 409A where the total does not exceed twice the lesser of the employee’s prior year annualized pay or the compensation limit for the year of termination, $360,000 for 2026, and where it is paid by the end of the second year following separation.

That safe harbor covers most small business severance comfortably. It stops covering it the moment somebody negotiates payments spread over four years, or a package well above the cap, and at that point an agreement drafted as an employment document has quietly become a deferred compensation plan.

When It Is Genuinely the Right Tool

There is a real use case and it is narrower than the sales material suggests: a stable, profitable business with a small number of senior people who are already maxing every qualified option and expect to be in a lower tax position when the money arrives.

Every element of that sentence carries weight. Stable, because the promise is unsecured and the recipient should discount it by their honest estimate of the odds. Profitable, because the deduction does not arrive until the employee is taxed, so you fund the liability out of after tax cash for years. Already maxing, because a plan that costs several thousand dollars to draft is absurd if the participant is deferring less than the qualified plan would have allowed anyway.

Two additional situations do genuinely qualify. A business heading toward a sale can use a transaction bonus tied to a defined change in control, which is a permitted payment event and aligns the incentive exactly. And a nonprofit employer faces a different statute altogether, section 457, where the tax-exempt version has its own rules and its own hazards.

One further advantage is worth knowing about because it is invisible until it matters. Federal law limits the ability of a state to tax the retirement income of a former resident, and payments from a nonqualified plan made in substantially equal installments over at least ten years fall inside that protection. For somebody who intends to retire out of a high tax state, that structure can be worth more than the deferral itself.

The last honest point is about who this is really for. A nonqualified plan is an owner’s tool far more often than it is an employee benefit, and there is nothing wrong with that as long as it is named accurately. If the reason for the plan is that you want to defer your own compensation, say so, and evaluate it as a personal tax decision rather than a retention program. If the reason is retention, the answer is almost always cash on a schedule, documented properly, paid before the middle of March.

Frequently Asked Questions