Late 401(k) deposits do not age out of Form 5500 filings
A late deposit is a dated event, so the instinct is to treat last year’s failure as last year’s filing problem. The Form 5500 instructions are built the other way: an uncorrected late deposit follows the plan onto every subsequent filing until it is actually fixed. This page quotes the rule, then shows what it looks like across 6,769 real filings.
The reporting rule: carried over until the year after full correction
From the 2024 Instructions for Form 5500, Schedule H line 4a (read 2026-08-25):
“The total amount of the delinquent contributions should be included on line 4a of the Schedule H or I, as applicable, for the year in which the contributions were delinquent and should be carried over and reported again on line 4a of the Schedule H or I, as applicable, for each subsequent year until the year after the violation has been fully corrected, which correction includes payment of the late contributions and reimbursement of the plan for lost earnings or profits.”
Two things follow. Line 4a is not a count of fresh failures each year: it is a running ledger, and an amount that reappears is an amount still open. And “fully corrected” is defined in the same sentence: principal paid, plus lost earnings. Depositing the contributions late but never funding the earnings leaves the item open, and the filing keeps saying so.
What the ledger behavior looks like in the data
We hold a cohort of 6,769 plans with 100 or more participants whose Schedule H line 4a said “Yes” in 2022, 2023 and 2024, all three years, built from the public EFAST2 datasets (method below). Inside that cohort, computed on 2026-08-25:
| observation | plans | share of cohort |
|---|---|---|
| identical positive amount reported in 2023 and 2024, to the cent | 1,638 | 24.2% |
| identical positive amount reported in all three years | 620 | 9.2% |
| amount strictly increasing across the three years | 1,558 | 23.0% |
An amount repeated to the cent across filings is exactly what the carry-over instruction produces when nothing has moved: the same open item, restated. A strictly increasing amount is the other shape of trouble: new delinquencies stacking on top of whatever came before.
The aggregate tells the same story. The cohort’s reported totals are $2,613,638,977 for 2022, $2,931,323,666 for 2023 and $2,926,967,318 for 2024, with medians of $50,819, $65,253 and $55,648 (computed on 2026-08-25 on the same cohort file). Three years of a population defined by uncorrected failures, and neither the total nor the median drains. If corrections were happening at scale inside this cohort, the 2024 numbers would be the smallest. They are not.
What an identical number can and cannot prove
Honesty about the instrument. A repeated amount is consistent with an uncorrected item carried forward per the instruction. It is also consistent with a sponsor copy-pasting last year’s filing without checking, which is itself a filing accuracy problem, just a different one. The public data cannot distinguish the two for any single plan. What it can say is distributional: one in four plans in this cohort shows the exact signature the carry-over rule stamps on an uncorrected failure, and the cohort’s totals behave like a pool that is filling, not draining.
One more thing the data cannot see: a plan that corrected mid-2024 still answers “Yes” for 2024, because the instruction runs “until the year after the violation has been fully corrected”. The 2025 filings will show who actually closed their items. This cohort is a floor on trouble, not a verdict on any one plan.
Why an open item is expensive to leave open
The cost side lives on a different form and is covered in depth in the cohort cost page. The short version, from the Form 5330 instructions (rev. December 2025): the section 4975 excise tax runs at 15% of the amount involved “for each year or part of a year in the taxable period”, and the taxable period ends at correction, not at year end. An item that sits open accrues filing years. The carry-over rule above is why we can see, from the outside, roughly how long items have been sitting.
Method
Computed on 2026-08-25 on chronic-cohort.csv, 6,769 plans of 100+ participants answering “Yes” on Schedule H line 4a in 2022, 2023 and 2024, built by pipeline.py from the public EFAST2 Form 5500 datasets (read 2026-08-21, re-verified 2026-08-25). The identical-amount and trend counts come from blog_stats.py in the same directory: a positive amount is counted as identical only on exact equality of the reported values. Amounts are the values sponsors themselves reported on line 4a; no estimation and no imputation. The full method is documented at planremit.com/method.
Published by Neige AI, Inc., last reviewed August 25, 2026. See the method and sources.
This is independent research, not legal or tax advice. It quotes primary sources with citations. Verify anything load-bearing against the primary source itself before acting on it.