Highly Compensated Employee (HCE)

A highly compensated employee is a classification used in retirement plan nondiscrimination testing. Under IRC §414(q)(1), an employee is highly compensated if they were a 5-percent owner at any time during the year or the preceding year, or if their compensation for the preceding year exceeded an indexed dollar threshold — $160,000 for 2026 — and, where the employer elects it, they were also in the top-paid group, meaning the top 20% of employees by compensation.

The classification exists for one reason: tax-favoured retirement plans are only allowed to be tax-favoured if they do not disproportionately benefit the people who own or run the business. To measure that, the rules need a defined group to compare against everyone else, and this is that group. Nothing about the label restricts what an individual may contribute.

The name is shared with two unrelated rules, so it is worth being explicit about which one is in play. This page is about IRC §414(q), the retirement plan meaning. The Department of Labor uses "highly compensated employee" for an entirely different purpose — an exemption from federal overtime requirements under 29 CFR §541.601, with its own salary test, its own duties test, and no connection to retirement plans. That threshold is set by regulation, and a 2024 rule raising it was vacated nationwide by a federal court in November 2024 — so the higher figure circulated widely and then stopped being the law, which is why the number should be checked at the Department of Labor rather than taken from a summary. Separately, IRC §105(h)(5) defines a "highly compensated individual" for self-insured medical plan testing using a third set of tests — the five highest-paid officers, a more-than-10% shareholder, or the top 25% of all employees.

The two prongs use different years. Calling this "a prior-year test" is the most repeated error about it, and it is only half right. The compensation prong is genuinely a look-back test: it asks what you earned in the preceding year, so a person hired mid-career at a large salary is generally not highly compensated in their first year and becomes one the following year. The ownership prong is not a look-back test — it asks whether you were a more-than-5% owner at any time during the current year or the preceding year. Someone who buys into a business in March is therefore a highly compensated employee for that same year, with no prior-year compensation involved at all.

The ownership prong has no dollar floor. This surprises small business owners regularly. A more-than-5% owner is highly compensated regardless of compensation — a founder taking no salary at all still counts. Note also that the test is "more than 5 percent," so exactly 5.0% is not a 5-percent owner, and that ownership is determined with the family attribution rules of IRC §318 in play. A spouse, child, grandchild, or parent can be attributed ownership and become a highly compensated employee without holding any direct stake, which is where family businesses most often get their testing wrong.

There is no officer prong. §414(q) does not make officers highly compensated. Officers appear in the separate definition of a key employee under IRC §416(i), which drives top-heavy plan testing. "Officers are automatically HCEs" is a real and common mistake that crosses two different tests with different thresholds and different consequences.

The top-paid-group election narrows the group and is sticky. An employer may elect to treat the compensation prong as reaching only employees who are also in the top 20% by compensation. The election must be in the plan document, must be applied consistently across the employer's plans, and generally cannot be changed without the Commissioner's consent. It can only ever make the highly compensated group smaller, and it cannot rescue a more-than-5% owner from the classification — the ownership prong is untouchable by the election.

What the classification actually does. Highly compensated employees are the measured group in the actual deferral percentage and actual contribution percentage tests, which compare their average deferral and matching percentages against everyone else's. If a test fails, the usual corrections are refunding a portion of the highly compensated participants' deferrals — a corrective distribution — or the employer making additional contributions for everyone else. That is a consequence of a failed test, not of the status: the ordinary contribution limit applies to a highly compensated employee exactly as it does to anyone else in the 401(k). A plan that avoids the tests altogether by adopting a safe harbor 401(k) design never reaches the question, and employers who want to give this group more room outside the qualified plan generally look to nonqualified deferred compensation.

A plan's administrator identifies the highly compensated group each year by running both prongs, then uses that split to test whether the plan is benefiting owners and top earners disproportionately.

A hypothetical example, showing why the years matter. Priya is hired on 1 March 2026 at a salary of $250,000, with no previous connection to the employer. For 2026 she had no prior-year compensation from this employer, so the compensation prong does not reach her and she is not a highly compensated employee that year — she can defer at whatever rate she likes without affecting the tested group. Her 2026 pay is well above $160,000, so for 2027 she is a highly compensated employee.

Dev's path is the opposite. On 1 June 2026 he buys 10% of the same company and draws a modest $40,000 salary. The ownership prong catches him immediately: he is a more-than-5% owner during 2026, so he is a highly compensated employee for 2026, with no compensation test applied and no prior-year history needed. If the company were owned by his spouse instead and he simply worked there, family attribution under §318 could still put him in the same position. Figures are illustrative.

Why the classification is useful

• It gives the nondiscrimination rules an objective, auditable definition of "the people who run the place" rather than a judgment call.
• Because the compensation prong looks back a year, plans can identify the group before the plan year starts and administer it prospectively.
• The top-paid-group election lets employers with many well-paid staff avoid sweeping most of the workforce into the tested group.

Where it causes problems

• The two prongs run on different years, so plans routinely misclassify new owners and newly hired executives in opposite directions.
• Family attribution catches relatives who own nothing and have no idea they are in the group.
• Highly compensated participants can find part of their deferrals refunded after year-end through a corrective distribution, which is an unwelcome and taxable surprise.
• The phrase is shared with unrelated overtime and medical plan tests, so searching for it produces confidently wrong answers.