Synthetic Form 941-SS - Employer's Quarterly Federal Tax Return (American Samoa, Guam, CNMI, USVI) Data
Synthetic training data — no real PII, fully coherent identities
Generate synthetic Form 941-SS quarterly employer returns for American Samoa, Guam, the Northern Mariana Islands, and the U.S. Virgin Islands. Visually near-identical to the mainland Form 941 but with the federal income tax withholding line removed — the classic near-duplicate classification trap.
144
Fields per document
6
Pages
Tax
Category
What this document is
Form 941-SS is the quarterly federal tax return filed by employers in American Samoa, Guam, the Commonwealth of the Northern Mariana Islands, and the U.S. Virgin Islands. It reports Social Security and Medicare taxes for territorial employees. Structurally it mirrors the mainland Form 941 — same parts, same deposit schedule logic, same signature block — with one decisive difference: there is no federal income tax withheld line, so the numbering jumps from line 2 straight to line 4.
Why generate synthetically
This is the best near-duplicate classification problem in the U.S. tax form corpus. A model that classifies documents by visual layout will confidently label a 941-SS as a 941, and then its extraction schema will look for a federal-withholding line that does not exist and either hallucinate a value or silently shift every subsequent line by one. Training a classifier to tell these two apart requires examples of both, and territorial filings are rare enough that a naturally collected corpus will contain almost none.
What makes synthetic data useful
The territorial employer population runs on its own, far smaller scale: the median 941-SS filer reports 3 compensated employees against 12 for the mainland 941 generated from the same identity corpus. Social Security and Medicare wage bases, the tax amounts derived from them, deposit schedule selection, and the monthly liability grid are generated as a consistent set, so line 5e and line 6 actually reconcile the way a reviewer would check.
Training challenges
The absence of the federal income tax line is the trap: everything above and below it looks like a 941, so a model must notice a missing row rather than recognize a present one — a much harder discrimination than adding a feature. The line 5 block splits Social Security wages, Social Security tips, Medicare wages, and Additional Medicare Tax wages into a column-1 base and a column-2 computed tax at four different rates stacked in adjacent rows, which is dense arithmetic in a small space. Deposit schedule selection branches the layout: monthly depositors fill a three-cell liability grid while semiweekly depositors attach Schedule B, and the corpus splits almost exactly evenly between the two. And the six-page PDF carries fillable content on only four pages, so page-level models must learn to skip instruction pages rather than hunt for fields on them.
Generate synthetic Form 941-SS - Employer's Quarterly Federal Tax Return (American Samoa, Guam, CNMI, USVI) data
Start with 500 free credits. No credit card required.
Generate NowWho uses this data
Payroll providers and tax software vendors serving employers in the U.S. territories, IDP teams whose document classifiers must separate near-identical IRS form variants before routing to an extraction schema, and compliance and audit groups processing historical territorial filings. If your classifier has never been shown a 941-SS, it is currently mislabeling every one it sees as a 941.
Document complexity profile
144 fields laid out across a 6-page PDF with only 4 field-bearing pages: 80 text fields, 43 currency amounts, 19 checkbox targets, 1 numeric counter, and 1 date. 170 annotation relations. The binding graph is computation-heavy for the size — 19 conditional bindings, 1 arithmetic binding, and 116 function calls at a maximum expression depth of 2 — because the quarter selector, the deposit schedule branch, and the wage-times-rate reconciliations all resolve at fill time.
Key stats from our synthetic corpus
Quantitative characteristics of the Form 941-SS - Employer's Quarterly Federal Tax Return (American Samoa, Guam, CNMI, USVI) documents our generator produces.
| Metric | Value | Detail |
|---|---|---|
| Median employees reported | 3 | The median territorial employer reports 3 compensated employees (p25 2, p75 5, max 21), against a median of 12 on the mainland Form 941 in the same corpus. The two forms draw employee counts through different generator paths, so the territorial corpus is not a re-badged mainland corpus. |
| Deposit schedule split | 49% / 51% | 49.5% of filings identify as monthly depositors and 50.5% as semiweekly depositors for part of the quarter. Because the two selections gate different layouts — a three-cell liability grid versus a Schedule B attachment — an even split is what lets a model learn both branches. |
| Quarter distribution | ~25% each | Filings spread across quarters at Q1 23.4%, Q2 26.4%, Q3 26.1%, and Q4 24.2%. The quarter selector is a four-checkbox radio group in the header and needs balanced coverage to train reliably. |
| Overpayment disposition | 35% apply / 16% refund | 35.4% of filings elect to apply an overpayment to the next return and 15.8% request a refund; the remainder have no overpayment. Seasonal employers account for 4.4% and closed businesses for 2.3%, two low-frequency checkboxes that a model will never learn from a small corpus. |
| Population coverage | 64% | 641 of 1,000 simulated identities are employed with an employer and therefore eligible to appear on a 941-SS, so a 1,000-identity run yields roughly 640 documents. |
How this document co-occurs with others
Rates at which identities in our corpus that produce a Form 941-SS - Employer's Quarterly Federal Tax Return (American Samoa, Guam, CNMI, USVI) also produce other documents.
| Correlation | Rate | Detail |
|---|---|---|
| Mainland near-duplicate | 100% | The mainland Form 941 is the form a classifier will confuse this one with. Generate both over the same identity set and you get a matched pair whose only structural difference is the missing federal withholding line — the cleanest possible discrimination training signal. |
| Correction pathway | 100% | Form 941-X carries an explicit checkbox for whether the return being corrected was a 941 or a 941-SS, so the two forms belong in the same dataset if you are training amendment workflows. |
| Employee wage statements | 100% | Every synthetic 941-SS employer issues W-2s. Quarterly Social Security and Medicare wages aggregate against the W-2 Box 3 and Box 5 totals for the same employer, giving a realistic annual reconciliation task. |
All stats above are corpus-derived: they were computed on a local synthetic corpus of 1,000 generated identities produced by SymageDocs' World Simulation Engine. No real employer or territorial payroll data was used. `make corpus-stats` regenerates the shipped corpus_stats.json for the site's analytics form set, which does not include this edition; reproduce these figures with `python -m symagedocs.corpus_analysis --corpus-size 1000 --form-id irs_f941ss_standard_2023`.
Frequently asked questions
- What data format do synthetic Form 941-SS documents include?
- Each generated identity produces a rendered PDF plus a structured JSON annotation file with bounding boxes, field types, and ground-truth values for all 144 fields across the four field-bearing pages of the six-page document. COCO, YOLO, FUNSD, and BIO/NER exports come from the same job.
- How is Form 941-SS different from Form 941?
- It omits the federal income tax withheld line entirely — territorial wages are not subject to federal income tax withholding — so line numbering runs 1, 2, 4, 5a and onward. The header names the four territories explicitly, and the Social Security and Medicare computation, adjustments, credits, deposit schedule, and signature sections otherwise parallel the mainland form. Generating both and diffing the annotations is the fastest way to see exactly where the layouts diverge.
- Which territories does it cover?
- American Samoa, Guam, the Commonwealth of the Northern Mariana Islands, and the U.S. Virgin Islands — all four named in the printed header. The form does not carry a per-territory selector; the jurisdiction is established by the employer address block.
- Can I use this data commercially?
- Yes. Every employer, wage figure, and tax amount is synthetic, contains no real payroll data, and is licensed for commercial use including model training, benchmarking, and redistribution inside your own products.
- Which revision year do you ship?
- The 2023 revision, which is the last 941-SS revision in our form library. Prior-year territorial returns keep circulating through amendment, audit, and recordkeeping workflows long after a revision stops being current, so an extraction pipeline that only knows the latest mainland form will keep meeting this layout for years.