Three years of “Yes” on line 4a, and what it actually costs

The number this page carries, and why nobody else has it

6,769 ERISA plans of 100+ participants answered “Yes” on Schedule H line 4a in 2022, 2023 AND 2024, all three years. Computed on the public EFAST2 filings, 2026-08-21, and re-verified 2026-08-25. Cohort file: staged/planremit/chronic-cohort.csv, built by staged/planremit/pipeline.py.

That single figure is the page. It is not a survey, not an estimate, and not a number any vendor publishes: it requires joining three filing years and keeping only the plans present in all three.

What the cohort actually looks like, and this is where it gets counterintuitive:

The headline is $8.5 billion and the reality is a mid-size plan with $220,000. The billions are carried by a short tail of very large filers, one of which reports $399 million on its own. Four out of five plans in this cohort are under a million dollars across three years, and nearly three out of four have fewer than a thousand participants. This is not a large-employer problem wearing a large-employer number.

Why “Yes” three years running is a specific fact, not a coincidence

The Schedule H instruction is explicit, and it is the reason a three-year cohort can be built at all:

“Continue to answer 'Yes' for any prior year failures until fully corrected.”

(Schedule H, Form 5500, line 4a instructions. Line 4a asks: "Was there a failure to transmit to the plan any participant contributions within the time period described in 29 CFR 2510.3-102?")

So a plan does not answer “Yes” three years running because it was late three separate times. It answers “Yes” three years running because the failure was still not fully corrected. The cohort is therefore not a list of repeat offenders. It is a list of open items.

That distinction matters for what follows, because the excise tax is priced on time.

The two ways sponsors get the cost wrong, in opposite directions

Too high: it is not 15% of the deposit

Section 4975(a) imposes a 15% tax, and the instinct is to apply it to the late contribution. The Form 5330 instructions (rev. December 2025) say otherwise:

“For purposes of calculating the excise tax on a prohibited transaction where there is a failure to transmit participant contributions (elective deferrals) or amounts that would have otherwise been payable to the participant in cash, the amount involved is based on interest on those elective deferrals.”

The 15% applies to the interest on the deferrals, not the deferrals themselves. A sponsor who has been quietly dreading a five-figure bill computed on a $220,000 figure is dreading the wrong arithmetic.

Too low: it is 15% per year, not once

Same instructions, and this is the sentence that gets skipped:

“The initial tax on a prohibited transaction is 15% of the amount involved in each prohibited transaction for each year or part of a year in the taxable period.”

And the taxable period:

“the period of time beginning with the date of the prohibited transaction and ending with the earliest of: (1) The date the correction is completed, (2) The date of the mailing of a notice of deficiency, or (3) The date on which the tax under section 4975(a) is assessed.”

The clock does not stop at the end of the plan year. It stops at correction. A plan whose line 4a has said “Yes” since 2022 and is still uncorrected is not looking at one 15% bite on each late deposit. It is looking at one for each year, or part of a year, that the transaction has sat uncorrected, and “part of a year” means a partial year counts as a whole one.

That is the actual shape of the exposure, and it is why the three-year cohort is the interesting population rather than the one-year one. The 6,769 plans above are, by their own filings, the ones where the multiplier has had the most time to run.

What this page does not tell you

Three things this page deliberately does not claim, because the public filings cannot support them:

1. Whether any specific plan owes anything. Line 4a is a yes/no about timing against 29 CFR 2510.3-102. It does not say whether lost earnings were deposited, whether a Form 5330 was filed, or whether the plan went through the DOL’s Voluntary Fiduciary Correction Program. A plan can answer “Yes” and be mid-correction.
2. The dollar amount of the excise tax for any plan. The amount involved is interest, and interest depends on deposit dates the filings do not carry. The $220,442 median above is the reported late contribution amount, never the tax.
3. Whether the sponsor knows. That is the only question worth asking a human, and it is the one this cohort exists to ask.

If you want to see whether your own plan is in the 6,769, the checker at planremit.com/check reads the same public filings by EIN and shows you the three years side by side. It reads what you already filed; it does not compute your excise tax, for the reason given above.

Method

Source: EFAST2 public Form 5500 datasets, read 2026-08-21. Selection: plans filing Schedule H with 100 or more participants that answered “Yes” on line 4a in each of the 2022, 2023 and 2024 filing years. Amounts are the values the sponsor itself reported on the same line; the three-year total is their sum, with no adjustment, no inflation, and no estimation. Plans present in fewer than three of the years are excluded rather than imputed. Full pipeline, including the exact field names and the exclusions: staged/planremit/pipeline.py.

Where a figure is a count of filings rather than a count of employers, it says so: a sponsor with two plans appears twice, because the obligation is per plan.