Data
Final retirement and benefit plan filings
Last updated October 24, 2026
Next update November 2026
An employer that ends its 401(k) or pension plan files one last annual return for it, with a box ticked: final return/report. The Department of Labor publishes those returns. We read the ones filed since January 2025 by employers whose plan had 25 or more active participants. 4,127 employers are left after our filters, with 240,869 active participants between them.
A final filing says a plan ended. It does not say the company closed. Plenty of employers end a plan because they were bought, merged plans, or switched providers. So we run a closure check on each one. Of the 1,389 checked so far, 81 read as closed, 5.8%. The rest of this page shows where the 4,127 sit and what the check has found.
By plan year and form
The plan year is the year the plan’s last return covers. The 5500-SF is the short form. The Department of Labor’s instructions say a plan filing it must “generally have fewer than 100 participants at the beginning of the plan year”.
| Plan year of the final filing | Form 5500-SF | Form 5500 | Filings |
|---|---|---|---|
| 2024 | 1,248 | 3 | 1,251 |
| 2025 | 2,356 | 1 | 2,357 |
| 2026 | 519 | 0 | 519 |
| All plan years | 4,123 | 4 | 4,127 |
Each employer counts once, at its latest kept filing. The filings were received from 2 January 2025 to 24 September 2026. 2,350 arrived in 2025 and 1,777 in 2026 to date.
By active participant band
The 5500-SF instructions describe active participants, for a pension plan, as people “currently in employment covered by the plan”. That includes staff eligible to put pay into a 401(k). We take the larger of the counts at the start and the end of the plan year. A plan that has wound up often reports zero at its end.
| Active participants | Filings | Share | Active participants, total |
|---|---|---|---|
| 25 to 49 | 2,625 | 63.6% | 89,992 |
| 50 to 99 | 1,136 | 27.5% | 76,226 |
| 100 to 249 | 302 | 7.3% | 40,662 |
| 250 or more | 64 | 1.6% | 33,989 |
| All bands | 4,127 | 100% | 240,869 |
Shares are of the 4,127, rounded to one decimal. The mean is 58.4 active participants per filing.
By plan year and band
| Plan year | 25 to 49 | 50 to 99 | 100 to 249 | 250 or more | Filings |
|---|---|---|---|---|---|
| 2024 | 793 | 355 | 86 | 17 | 1,251 |
| 2025 | 1,508 | 636 | 178 | 35 | 2,357 |
| 2026 | 324 | 145 | 38 | 12 | 519 |
What our closure check found
The check is one web search per employer, read for wording about a shutdown, a sale or a site closing. The table gives each employer’s latest verdict.
| Active participants | Filings | Closed | Acquired | Site closure, company continues | Unclear | Not checked yet |
|---|---|---|---|---|---|---|
| 25 to 49 | 2,625 | 0 | 0 | 0 | 1 | 2,624 |
| 50 to 99 | 1,136 | 61 | 363 | 106 | 504 | 102 |
| 100 to 249 | 302 | 15 | 107 | 33 | 140 | 7 |
| 250 or more | 64 | 5 | 20 | 5 | 29 | 5 |
| All bands | 4,127 | 81 | 490 | 144 | 674 | 2,738 |
The same verdicts by plan year:
| Plan year | Filings | Checked | Closed | Acquired | Site closure | Unclear | Not checked yet |
|---|---|---|---|---|---|---|---|
| 2024 | 1,251 | 442 | 33 | 164 | 41 | 204 | 809 |
| 2025 | 2,357 | 783 | 39 | 270 | 81 | 393 | 1,574 |
| 2026 | 519 | 164 | 9 | 56 | 22 | 77 | 355 |
| All years | 4,127 | 1,389 | 81 | 490 | 144 | 674 | 2,738 |
Of the 1,389 checked, 5.8% read as closed, 35.3% as acquired, 10.4% as a site closure of a company that continues, and 48.5% as unclear.
By state
The state is the one on the plan sponsor’s mailing address. That’s usually the head office, not every place the employer had staff.
| State | Filings | Active participants |
|---|---|---|
| CA | 593 | 33,448 |
| TX | 268 | 24,343 |
| NY | 241 | 14,832 |
| FL | 225 | 12,969 |
| OH | 209 | 10,581 |
| IL | 183 | 11,592 |
| PA | 169 | 10,531 |
| GA | 131 | 6,349 |
| NJ | 131 | 7,417 |
| VA | 130 | 7,661 |
| IN | 113 | 5,483 |
| NC | 112 | 7,168 |
| MI | 109 | 6,177 |
| MO | 104 | 5,059 |
| MA | 101 | 5,349 |
| MN | 90 | 4,649 |
| WA | 84 | 5,114 |
| AZ | 79 | 6,557 |
| CO | 78 | 4,325 |
| MD | 78 | 3,975 |
| WI | 73 | 3,817 |
| TN | 72 | 3,885 |
| OR | 53 | 2,817 |
| CT | 52 | 2,841 |
| KY | 51 | 2,209 |
| SC | 50 | 2,581 |
| LA | 46 | 2,400 |
| KS | 43 | 2,459 |
| AL | 39 | 2,032 |
| NV | 38 | 2,078 |
| IA | 34 | 1,804 |
| UT | 33 | 1,677 |
| OK | 28 | 2,395 |
| NH | 26 | 1,379 |
| AR | 22 | 1,003 |
| ME | 22 | 1,223 |
| NE | 22 | 966 |
| ID | 21 | 1,167 |
| RI | 20 | 846 |
| WV | 18 | 887 |
| MS | 17 | 974 |
| NM | 17 | 805 |
| VT | 16 | 985 |
| DE | 15 | 694 |
| MT | 15 | 826 |
| DC | 13 | 488 |
| ND | 9 | 424 |
| HI | 8 | 405 |
| SD | 8 | 360 |
| AK | 7 | 275 |
| WY | 5 | 262 |
| PR | 3 | 212 |
| GU | 2 | 84 |
| No state given | 1 | 30 |
| All | 4,127 | 240,869 |
What the tables show
The list is mostly small plans. 63.6% of the filings name 25 to 49 active participants. Only 64 name 250 or more. Almost all came on the short form: 4,123 of 4,127.
The four largest states hold 32.2% of the filings: California with 593, Texas with 268, New York with 241 and Florida with 225. That roughly follows where employers are. Texas stands out on size. Its 268 filings carry 24,343 active participants, about 91 per filing, against about 56 in California.
The closure check tells a different story. Among the employers it has reached, about one in three reads as acquired. About one in two can’t be called either way. Closed is the smallest verdict: 81 employers, 5.8% of those checked. That’s close to what a first sample checked by hand on 23 October 2026 found: of 102 entities read by a person, 9 closed, 36 acquired, 9 site closures and 48 unclear.
The share that reads as closed barely moves with size: 5.9% of the checked 50-to-99 band, 5.1% of the 100-to-249 band, 8.5% of the 250-or-more band. The last rests on 59 checked employers, so read it loosely.
The check has barely started on the 25-to-49 band. One employer of 2,625 has a verdict. That band is most of the list, so the next update may move these shares.
Read the table as a list of plans that ended, not a list of closures. The filing exists because the plan’s money had to go somewhere. It says nothing about the payroll, the office or the books.
What this means for a closed company
If yours is among the 81, or among the ones nobody has checked yet, the records are still yours to decide on. A company that ran a plan for 40 people ran payroll, hiring and a dozen recurring processes. The plan’s end is often one paperwork step in a longer wind-down.
The records that matter to a buyer are the ones that show how the work got done. The Slack or Teams workspace. The shared mailboxes. The SOPs someone wrote because new hires kept asking. The project history, the support tickets, the CRM with its notes on every deal. micro1 says it is looking for “established companies with 30+ full-time employees, mature business operations, and substantial historical operational data.” It lists “internal communications, SOPs, project histories, CRM records, support tickets” among what that can include.
micro1 runs a separate track for companies like these. Its referral page names a “Defunct Company Data Partnership” and says: “We are acquiring data from companies that are no longer operating.” The company requirement it lists for that track is “30+ employees”. Our own line is 25: we refer a closed company to micro1 when it had 25 or more people at its peak. micro1 decides at application. It publishes no minimum number of years; its payout page asks “How many years of data do you have?” and adds “Typically the number of years you’ve been in business”. A buyer licenses records like these to train AI models.
For the steps, what the founder keeps and who can sign once the company is gone, see our page for closed companies. What the records can be worth and which accounts to export before a bill is cut is on what a closing company’s records are worth. The record types a buyer asks for are in our explainer on what data buyers actually want. Who can sign for a dissolved company is a question for your counsel.
Other public records catch other closures. Our table of WARN notices by state counts the larger employers that gave notice before a closing or layoff. Our count of company dissolutions in two state registries shows how long the companies that formally dissolved in Colorado and Connecticut had run.
If your company ended its plan and closed, our estimate asks five questions and gives you a band. It’s ours, not an offer. Do you still have admin access to the mailboxes?
What is never included
On this page: counts only. No company names, no employer identification numbers, no plan names, no signer names, no addresses, no phone numbers. The Department of Labor publishes those; we don’t repeat them.
In a licence: micro1 says “Companies retain ownership of their underlying data.” Its payout page says “Personal and identifying information is removed or transformed before your data is shared with third parties”. As our own advice: keep payroll, plan account statements, employee medical records, customer payment details and anything a client owns out of any export. Retirement plan records are about people; they’re not the operating record a buyer wants. How it’s safe sets out what leaves your systems and when, and our page on how de-identification works explains the steps.
Methodology
What is counted. Each number counts employers in our database that our Form 5500 collector created, one row per employer identification number. We read the host database with read-only queries on 24 October 2026; the figures here are from the read at 17:05 UTC. 44 further companies that we first recorded from another source also carry a final filing. They’re not in the tables.
The source. The U.S. Department of Labor’s Employee Benefits Security Administration publishes the Form 5500 datasets. Its page says they hold structured data from the Form 5500 and Form 5500-SF and their schedules, filed each year, and that the 2009 and later datasets are updated around the first of each month. We read the “Latest” files for plan years 2024, 2025 and 2026: the Form 5500 file, the Form 5500-SF file, and the lists of plans whose assets moved to another plan (Schedule H and Schedule I line 5b, and the 5500-SF line 13c).
What a final filing is. The 5500-SF instructions say to tick the final return/report box “if this is the final report for the plan”. The box is ticked only once all plan assets have been paid out to participants or moved to another plan. So the filing marks the end of a plan, whatever the reason.
What is kept. A filing stays in when it is marked final, its plan year begins on 1 January 2024 or later, it names a sponsor and an employer identification number, and the plan had 25 or more active participants at the start or the end of the plan year (the larger of the two counts). The latest filing per employer identification number wins, by date received.
What was left out, and why.
- Multiemployer plans (union funds) and multiple-employer plans (pooled plans, such as a professional employer organisation’s). One employer’s end says little about either.
- Plans whose assets moved to another plan, read from the transfer lists above. Those plans merged into another plan, which can be a buyer’s.
- Employers that file a non-final return for another plan with a plan year from the final one’s or later. An employer that ends a frozen pension and keeps its 401(k) is still trading.
- Plans with fewer than 25 active participants.
Plan types. We don’t filter by type of plan. A rough reading of the plan names gives about 90% retirement plans: 3,726 of the 4,127 (90.3%) contain 401, 403(b), retirement, pension, profit sharing, savings, thrift, cash balance, defined benefit, money purchase, SIMPLE IRA, ESOP, stock ownership or deferred comp. Another 169 (4.1%) contain none of those but one of health, welfare, medical, dental, vision, cafeteria, insurance, life, disability, benefit plan or benefits plan. 232 (5.6%) contain neither set. A word match is a reading of the name, not of the plan.
The form split. 4 of the 4,127 rows came from the full Form 5500. The short form is for plans that generally have fewer than 100 participants at the start of the plan year. Even so, 217 short-form filings report more than 120 active participants. The counts here are as filed, so read the top bands as the figures the sponsors reported. That’s also fewer full-form filings than the size bands would lead you to expect. Our likely reading, still to be checked, is the continuing-plan filter: an employer that files a non-final return for another plan is dropped, and larger employers usually keep a welfare or other plan running. The next update will report what we find.
Size. The band is the larger of the active participant counts at the start and end of the last plan year. It counts people in the plan, not staff: employees outside the plan aren’t in it. For a company that closed, the figure is a reading at one point in the plan’s last year, not the company’s peak.
The closure check. For each employer, we run one web search for its name with terms such as “shut down”, “ceased operations”, “out of business”, bankruptcy and acquired. The results that name the company are sorted by rule. Closed needs shutdown wording in at least two results from at least two different outlets, and no sign of a sale or a continuing company. Acquired means wording such as “acquired by” or “merged with”. Site closure means a closing at one place, or wording that the company carries on. Anything else is unclear. A verdict is a reading of what the web says, not proof either way. Each employer’s latest verdict is the one counted. Until the check reads an employer as closed, or a person clears it, we don’t contact it.
What the check has reached. Checks ran from 16 October to 24 October 2026; the latest was at 13:12 UTC on 24 October. Five verdicts predate the Form 5500 rows: they come from a check of the same company under another public record. Four rows carry no company key and can’t be checked.
Other exclusions downstream. 66 rows were set aside by our name screen, which keeps out hospitals, schools, universities, public bodies, staffing firms and recruiters. They are still counted in the tables above.
Refresh. The tables are rebuilt monthly from a new read of the database, after the datasets’ monthly update. The next update is due 24 November 2026.
TrustPast
Apply & License Your Data*Purchases are made by micro1 under its data partnership programme. TrustPast is an independent referral partner of micro1.
Sources
- U.S. Department of Labor, EBSA: Form 5500 datasets (Form 5500 and 5500-SF filings, updated monthly)
- U.S. Department of Labor, EBSA: 2024 Instructions for Form 5500-SF (who may file, the final return/report box, active participants)
- micro1, refer a company: the Defunct Company Data Partnership, company requirements 30+ employees (micro1.ai)
- micro1, data partnerships: the data it looks for, ownership, de-identification (micro1.ai)
- micro1, how much is your data worth: the years question, de-identification before sharing (micro1.ai)