Safe Harbor 401(k): How It Works and What It Costs
A safe harbor 401(k) skips nondiscrimination testing in exchange for a mandatory vested employer contribution. The formulas, deadlines, and trade-offs.
Most explanations of safe harbor 401(k) plans describe them as the better version of a 401(k). They are not better, they are a trade, and whether the trade is worth taking depends on a number most small business owners have never calculated.
The trade is this: you commit to an employer contribution that is mandatory and immediately vested, and in exchange the plan stops failing the annual tests that limit how much you and your highly paid people can defer. If those tests were never binding on you, you have bought an expensive solution to a problem you did not have.
This covers what the design actually does, the three formulas and what each costs at your participation rate, the deadlines that arrive earlier than people expect, and the situations where a plain 401(k) is the better answer.
I build the people and records tooling for businesses without an HR department at FirstHR, and FirstHR is an onboarding and HR platform rather than a retirement plan provider. This is general information, not tax or investment advice.
What a Safe Harbor Plan Is
A safe harbor 401(k) is an ordinary 401(k) with a specific set of provisions that take it outside the annual deferral and matching nondiscrimination tests.
According to the IRS, a safe harbor 401(k) is not subject to the complex annual nondiscrimination tests that apply to traditional 401(k) plans, and it must provide for employer contributions that are fully vested when made (Internal Revenue Service). The underlying framework sits in the retirement plan qualification rules (26 U.S.C. 401).
The Problem It Solves
An ordinary 401(k) is tested annually to check that highly compensated employees are not deferring disproportionately more than everybody else. When they are, the plan fails. The usual correction is unpleasant: contributions get refunded to the highly compensated employees, taxable to them, and for a calendar year plan that usually happens by mid-March (26 CFR 1.401(k)-2).
At a small business this failure is almost structural. If the owner and two senior people defer heavily and eight hourly employees defer nothing, the averages will not work, however generous the plan design looks on paper. The safe harbor exists precisely for that shape of business, and the conditions it has to meet are set out in the regulations (26 CFR 1.401(k)-3).
The first question is therefore not how to set one up. It is whether your plan is actually failing, or close to failing, because a plan that passes comfortably gains nothing from the trade and pays for it anyway.
The Three Formulas
Leaving aside the automatic enrollment version, three designs satisfy the safe harbor, and they behave very differently against a real payroll.
The enhanced match exists mostly because the basic formula is difficult to explain. One hundred percent of the first four percent is a sentence anybody understands; one hundred percent of the first three plus fifty percent of the next two is a sentence people nod at without following.
Both cap at four percent of pay, and the enhanced version costs more only for people deferring between three and five percent.
What the Safe Harbor Match Pays in Practice
Run the safe harbor match against one salary and the formulas stop being abstract. Take an employee earning $60,000 who defers 4 percent of pay, or $2,400. The basic formula matches $1,800 on the first 3 percent, then $300 on the next 1 percent, for $2,100 in total.
The enhanced formula matches the whole first 4 percent, so the same person receives $2,400, a $300 gap on one salary. Push the deferral to 5 percent and both formulas land on $2,400, since each caps at 4 percent of pay. Anybody deferring nothing receives nothing under either.
Match or Nonelective?
Which one is cheaper depends on participation: a match wins when few people defer, and a nonelective often wins when nearly everyone does. This is the decision with real money in it, and the answer runs opposite to most people's instinct.
| Scenario | Match is cheaper | Nonelective is cheaper |
|---|---|---|
| Participation around 30 percent | Yes, substantially | No |
| Participation around 60 percent | Usually | Approaching parity |
| Participation above 85 percent | Often not | Frequently yes |
| Cost predictability year to year | Low, it varies with participation | High, it is a fixed percentage of payroll |
| Employees who defer nothing | Cost you nothing | Still receive 3 percent |
| Administrative notice burden | Annual notice required | No annual notice required |
The fifth row is the one that decides most small business cases, and it cuts both ways. A match rewards people who participate and gives nothing to those who do not, which is either fair or a missed opportunity depending on your view.
A nonelective gives everybody three percent whether they engage or not. That costs more when participation is low, and it is genuinely valued by lower-paid employees who cannot afford to defer.
The fourth row matters more than it looks for a business with variable revenue. A nonelective is a known percentage of payroll you can budget once. A match is a number that moves when your people change how much they defer, which makes it harder to forecast in exactly the years you most need to.
None of that answers the question for your payroll, and the only thing that does is running both formulas against your own eligible compensation and your own deferral rates. One row per eligible employee and three cost columns is the whole exercise.
| A | B | C | D | E | F | G | |
|---|---|---|---|---|---|---|---|
| 1 | Employee or ID | Eligible compensation | Deferral rate | Basic match cost | Enhanced match cost | Nonelective cost | Notes |
| 2 | SAMPLE owner | Highly compensated, defers the maximum | |||||
| 3 | SAMPLE hourly staff | 0% | Defers nothing, still receives the nonelective | ||||
| 4 | |||||||
| 5 | |||||||
| 6 | |||||||
| 7 | |||||||
| 8 | |||||||
| 9 | |||||||
| 10 | |||||||
| 11 | |||||||
| 12 | |||||||
| 13 |
What You Give Up
Four commitments come with the design, and the first is the one employers object to most.
Immediate vesting is a genuine loss for businesses that were using a graded vesting schedule as a retention mechanism. Traditional safe harbor contributions cannot be subject to one, so somebody who leaves after nine months takes the whole employer contribution with them.
Automatic enrollment designs are the one exception. A qualified automatic contribution arrangement may require up to two years of service before its safe harbor contributions become nonforfeitable (Internal Revenue Service), the legal term for vested.
That service requirement of up to two years is the only vesting schedule available anywhere inside the design. It comes attached to automatic enrollment, so it is not a route to a schedule on its own.
Whether that matters is worth being honest about. Vesting schedules retain people who were already ambivalent, and their effect is usually smaller than employers believe. But if you were relying on one, this design removes it.
The Deadlines
A new plan using a safe harbor match has to run for a first plan year of at least three months, which for a calendar year plan puts the practical deadline at October 1 (26 CFR 1.401(k)-3). The single most common way a small business misses a safe harbor year is by starting the conversation in November.
An existing plan runs on a different clock, and a later one. The safe harbor provisions have to be adopted before the first day of the plan year, and the notice is timely when it reaches eligible employees between thirty and ninety days before that day, so December 1 is the date that actually binds a calendar year plan.
The annual notice, where required, runs on its own clock. The IRS treats it as timely when it arrives at least thirty and not more than ninety days before the plan year begins, or, for somebody who becomes eligible mid-year, no later than the eligibility date (Internal Revenue Service).
Missing a required notice is a plan failure rather than an administrative slip: the IRS classes it as an operational failure that has to be corrected (IRS correction guidance). That is one more reason match designs are slightly riskier to run than nonelective ones.
Nonelective designs no longer carry the annual notice. According to IRS Notice 2020-86, the SECURE Act removed the safe harbor notice requirement for them, automatic enrollment versions included, for plan years beginning after December 31, 2019. That leaves the notice on match designs.
What It Actually Costs
The contribution is the visible cost, up to 4 percent of pay for a match or 3 percent of eligible pay for a nonelective, and it is not the only one.
| Cost | Match design | Nonelective design |
|---|---|---|
| Employer contribution | Up to 4 percent of pay, only for participants | 3 percent of all eligible compensation |
| Predictability | Varies with participation | Fixed percentage of payroll |
| Payroll tax on contributions | None, employer retirement contributions are not wages | None |
| Administration and recordkeeping | Provider fees, generally similar either way | Similar, with one fewer notice to produce |
| Annual notice production | Required | Not required |
| Cost of not doing it | Corrective distributions to owners and highly paid staff | Same |
That last row is what the whole decision turns on. If your plan currently fails testing, the alternative to a safe harbor contribution is refunding money to exactly the people who most wanted to save it, every year, taxable to them. Compared with that, a fixed contribution to everybody is frequently the better arrangement even before considering what employees think of it.
The payroll tax line in that table is worth stating plainly, because employers routinely assume otherwise. Employer contributions paid into a qualified plan trust sit outside the definition of wages for Social Security and Medicare (26 U.S.C. 3121).
The federal unemployment tax definition of wages excludes them on the same terms (26 U.S.C. 3306). A dollar of safe harbor contribution costs a dollar, which is not true of a dollar of bonus.
That does not make it a small addition to your benefits spend. Three percent of eligible payroll is a real number, and it sits alongside the rest of what you pay for benefits per employee rather than instead of any of it.
Dwight SchruteSales3 people
Darryl PhilbinWarehouse1 person
Kelly KapoorCustomer Service1 personWhen Not to Do It
Five situations where the honest answer is that a plain 401(k) is better, set against the situations where a safe harbor earns its cost.
The last item under Cons is the most common reason small businesses adopt one, and it is the worst reason. Ask your provider for the actual test results before deciding, because the answer is either a number that justifies the cost or a number that says you are fine.
Setting One Up
Most of the work is decisions rather than paperwork, and the paperwork is your provider's job.
Where Small Employers Get This Wrong
Five patterns, and the first is the expensive one.
Adopting it without checking whether testing was actually a problem is first. It is a real annual cost bought to solve a constraint that may not exist for you.
Assuming a match is always cheaper is second. At high participation the nonelective design is frequently cheaper and always more predictable.
Starting the conversation in November is third. Match designs need to be in place well before the plan year, and the nonelective escape route is not always available.
Treating the contribution as discretionary is fourth. Once adopted for a year it is a commitment, and exiting mid-year is possible only narrowly and returns you to full testing for that year.
And forgetting the annual notice is last, which is a plan failure rather than an oversight, and is the specific administrative reason some employers prefer the nonelective design.
Nick Anisimov, FirstHR founder
Frequently Asked Questions
What is a safe harbor 401(k)?
A safe harbor 401(k) is a retirement plan design that trades a required employer contribution for relief from the yearly nondiscrimination tests. The contribution, either a set matching formula or a flat percentage of pay for every eligible employee, belongs to the employee from the moment it is made. In return, the plan sits outside the annual deferral and matching tests that otherwise cap what owners and highly paid staff can put in. Small businesses use it most when the owners want to save meaningfully and too few other employees participate for an ordinary plan to pass testing. You accept a guaranteed cost in exchange for removing a constraint.
What are the safe harbor contribution formulas?
Three designs are common. The basic match adds a dollar for every dollar an employee defers up to 3 percent of pay, then 50 cents per dollar on deferrals between 3 and 5 percent, so it tops out at 4 percent of pay. The enhanced match is often a dollar for dollar match on the first 4 percent deferred, and it can never pay less than the basic formula at any deferral rate. The third option skips matching altogether: you contribute 3 percent of pay for every eligible employee, whether or not they save anything. In a traditional safe harbor, all three vest immediately.
Is a match or a nonelective contribution cheaper?
It depends entirely on participation, and the answer is counterintuitive. A match costs you only for employees who defer, so with low participation it is cheap. A nonelective costs 3 percent of all eligible compensation whether people participate or not, so with low participation it is expensive. As participation rises the two converge and then cross. Model both against your actual payroll and your actual deferral rates rather than assuming the match is always cheaper, because at high participation it frequently is not.
What are the disadvantages of a safe harbor 401(k)?
Three, and they are real. The contribution is mandatory once adopted for the year, so it becomes a fixed cost rather than a discretionary one you can suspend in a difficult quarter. It vests immediately, which removes the ability to use a vesting schedule as a retention tool. And it costs money you were not previously spending, which is only worth it if the testing constraint it removes was actually binding on you. A plan that passes testing comfortably gains little from the trade.
Does a safe harbor 401(k) still need nondiscrimination testing?
A safe harbor plan sits outside the annual deferral and matching tests, which is the entire point of it. Other requirements do not disappear. Top-heavy status has its own rules, and a plan that adds discretionary profit sharing on top of the safe harbor contribution can find itself back inside obligations it thought it had escaped. Coverage requirements also continue to apply. Confirm with your provider what your specific design does and does not remove rather than assuming it removes everything.
When is the deadline to set up a safe harbor 401(k)?
Earlier than most employers expect, and the exact date depends on the design. A brand-new plan that uses a safe harbor match needs a first plan year of at least three months, so for a calendar year plan October 1 is the practical cutoff. If you already have a plan, the safe harbor provisions must be in place before the plan year opens, and a match design also needs its notice delivered in the window the IRS treats as timely, thirty to ninety days beforehand, which makes December 1 the date that binds. A nonelective contribution is more forgiving. You can add it by amendment before the thirtieth day ahead of the plan year end, or before the last day of the following plan year if you raise it to at least 4 percent, as long as the plan offered no safe harbor match that year. Ask your provider for the exact dates in the summer rather than in November.
Do you have to give employees a notice?
For match-based designs, yes. Every eligible employee needs a written notice each year that explains their rights and obligations under the plan, and the IRS treats it as timely when it goes out thirty to ninety days before the plan year starts. Someone who becomes eligible partway through can receive it up to ninety days before eligibility, as long as it arrives by the date they become eligible. For plan years beginning after December 31, 2019, a plan that meets the safe harbor with a nonelective contribution no longer has to send one, a quiet administrative advantage of that design. A required notice that is missed counts as a plan failure that has to be corrected, not a formality.
Can you stop a safe harbor contribution mid-year?
Only in defined circumstances and never casually. The regulations allow a mid-year reduction or suspension where the employer is operating at an economic loss, or where the annual notice already warned that the contribution might be reduced during the year. Either route requires a supplemental notice, and the change takes effect no earlier than thirty days after employees receive it. The plan then has to satisfy the deferral test for the whole plan year on the current year method, which can produce exactly the corrective distributions the safe harbor was adopted to avoid. Budget the contribution as a fixed annual commitment, because the exit is expensive and awkward.
Make HR the easy part of running your business
Onboarding, e-signatures, training and core HR in one place. Up and running in 5 minutes.
- 7-day free trial
- No credit card required
- Cancel anytime