Derived table · PPP and COVID EIDL
PPP/EIDL identity-theft classification
1,977 PPP and EIDL prosecutions from the DOJ index, each classified as identity theft (246 cases, where another person's identity or information was used) or self-inflation (1,731 cases, where the defendant applied in their own name and misstated payroll, revenue, or documents), with the basis, a confidence grade, and a supporting excerpt.
Rows1,977
Columns22
Files1 · 1.0 MB
FormatCSV
Source vintageProject classification through 2026-07-10
Published byPandemic Darlings
Acquired / built2026-07-10
Releasepandemic-darlings-data-v1-2026-07-10
Built fromDOJ press-release corpus, 2020–2026
SHA-2564e7f846e17377d6d049caed33fd4c48f182875525083eec050e58a1630925bf1
RightsCompiled from public records; attribution to Pandemic Darlings requested for the classification work.
Schema · dictionaryppp-eidl-identity-theft-classification.json · dictionary.csv
Download CSV 1.0 MBData dictionary CSVAt a glancePreviewColumnsHow to useBefore you quote a row
At a glance
Measured on the served file on 2026-10-02; the checksum above is the file these numbers describe.
Rows1,977
Columns22
Key
uuid · uniqueCells left blank10,759 of 43,494 · 24.7%
Date range2020-05-13 → 2026-05-12
Identity theft: yes246 · 12.4%
Classified with full text661 rows
Reliable dollar figures1,644 rows
ppp-eidl-identity-theft-classification.csv by the year of date and the value of identity_theft.The numbers behind this chart
| Period | Identity theft: yes | Self-inflation: no | Total |
|---|---|---|---|
| 2020 | 16 | 122 | 138 |
| 2021 | 48 | 236 | 284 |
| 2022 | 46 | 315 | 361 |
| 2023 | 56 | 333 | 389 |
| 2024 | 42 | 334 | 376 |
| 2025 | 28 | 293 | 321 |
| 2026 | 10 | 98 | 108 |
ppp-eidl-identity-theft-classification.csv by classification_basis; the eight largest of 17 distinct values are shown.The numbers behind this chart
| Basis | Rows |
|---|---|
| no identity-theft charge or third-party-identity language | 2K |
| charge: aggravated identity theft (18 USC 1028A) | 173 |
| charge: identity theft | 19 |
| text:identities | 17 |
| text:x_of_others | 10 |
| hand-review:self-application (trigger false-positive) | 7 |
| text:without_consent | 6 |
| verified by full-text fetch (sample QA) | 4 |
What it is
Rows are the PPP and EIDL prosecutions from the DOJ press-release index, minus advisories and multi-case roundups. identity_theft is yes when the release describes a third party's identity or personal information being used: stolen or synthetic identities, applications in other people's names, deceased victims, family members used without their knowledge, or an aggravated-identity-theft charge, which by statute requires a means of identification of another person. It is no when the defendant applied in their own name or through their own real or invented entity and misstated payroll, employees, revenue, or tax documents.
The distinction matters for how detectable the fraud was. A self-inflation case can be checked against IRS records for the same entity; an identity-theft case cannot be caught that way. 459 of the self-inflation rows carry explicit evidence of inflated payroll or revenue in the text.
How it was built
Classification runs in three passes. A charge of identity theft or aggravated identity theft makes the row a yes at high confidence. Otherwise the full press-release body, or the summary where the body was unavailable, is matched against a tight set of third-party-identity phrases; every text match was reviewed by hand, and false positives (in the name of his business, tax preparers falsifying their own clients' applications) were overridden. Everything else is a no, with confidence set by how much text was available: high with the full body, medium with only the summary.
Preview
| uuid | date | title | defendant | district | programs | dollar_amount | dollar_amount_fmt | dollar_amount_reliable | charges | case_type | sentence | restitution_fmt | identity_theft | classification_basis | id_theft_subtype_note | confidence | self_inflation_evidence | supporting_excerpt | has_full_text | url | fact_source |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| d780f989-009c-4fdf-bfc9-5a0391896db5 | 2026-05-12 | Cheshire Man Who Defrauded Pandemic Relief Program is Sente… | Cheshire Man Who Defrauded Pandemic Relief Program is | CT | EIDL | 96200.0 | $96,200 | yes | theft of government money, making an illegal monetary trans… | sentenced | 3 months (0.2 yrs) | no | no identity-theft charge or third-party-identity language | high | yes | https://www.justice.gov/usao-ct/pr/cheshire-man-who-defraud… | canonical | ||||
| 6fd7dd81-5b36-43f2-941d-7c6192506032 | 2026-05-12 | Beverage Dispensing Equipment Manufacturer Pays $2.6 Millio… | MDFL | PAYCHECK PROTECTION PROGRAM, PPP | 2593219.18 | $2,593,219 | yes | false claims | settle | no | no identity-theft charge or third-party-identity language | medium | https://www.justice.gov/usao-mdfl/pr/beverage-dispensing-eq… | parsed | |||||||
| ea82c441-af86-4d6f-b2a1-2bb5a028645a | 2026-05-11 | Apopka Man Sentenced to Federal Prison for COVID-19 Fraud | Apopka Man | MDFL | PPP | 739582.0 | $739,582 | yes | wire fraud and money laundering | sentenced | 63 months (5.2 yrs) | no | no identity-theft charge or third-party-identity language | high | yes | of could qualify for forgiveness.As part of his application… | yes | https://www.justice.gov/usao-mdfl/pr/apopka-man-sentenced-f… | canonical | ||
| 1f889b05-33d6-4b53-89af-2cb349633e74 | 2026-05-07 | South Florida Tax Preparer Admits to Wire Fraud in $4.1 Mi… | SDFL | PAYCHECK PROTECTION PROGRAM, PPP | 4099999.9999999995 | $4,100,000 | yes | tax fraud, conspiracy, wire fraud | no | no identity-theft charge or third-party-identity language | medium | yes | MIAMI – A federal judge accepted the guilty plea of a South… | https://www.justice.gov/usao-sdfl/pr/south-florida-tax-prep… | parsed | ||||||
| 5a07f4dc-05c6-47b4-8e53-71fd461654e0 | 2026-05-07 | Brooklyn Clothing Company Pays $3 Million to Settle Claims … | EDNY | PAYCHECK PROTECTION PROGRAM, PPP | 3000000.0 | $3,000,000 | yes | false claims | settle | no | no identity-theft charge or third-party-identity language | medium | https://www.justice.gov/usao-edny/pr/brooklyn-clothing-comp… | parsed | |||||||
| 64b9fafb-af74-4c57-befd-80101f98af1e | 2026-05-06 | Marrero Woman Guilty of Tax Fraud and PPP Fraud | EDLA | PAYCHECK PROTECTION PROGRAM, PPP | 75000000.0 | $75,000,000 | yes | false statements | no | no identity-theft charge or third-party-identity language | medium | https://www.justice.gov/usao-edla/pr/marrero-woman-guilty-t… | parsed | ||||||||
| 13165f0e-b46c-4e99-954c-5b68c3b40bfe | 2026-05-06 | Shelby County Man Charged in Multi-Million Dollar COVID-19 … | Shelby County Man | NDAL | PAYCHECK PROTECTION PROGRAM, PPP, ECONOMIC INJURY DISASTER … | 3000000.0 | $3,000,000 | yes | wire fraud, aggravated identity theft | charged | yes | charge: aggravated identity theft (18 USC 1028A) | high | https://www.justice.gov/usao-ndal/pr/shelby-county-man-char… | parsed | ||||||
| 784e583b-e830-4a12-8ab0-f74cf7db3fb2 | 2026-05-05 | California Tax Preparer Pleads Guilty to Filing False Retur… | California Tax Preparer | PAYCHECK PROTECTION PROGRAM, PPP, EIDL | 73000000.0 | $73,000,000 | yes | wire fraud | pleaded guilty | no | no identity-theft charge or third-party-identity language | medium | https://www.justice.gov/opa/pr/california-tax-preparer-plea… | parsed | |||||||
| a0ae3ded-5c2a-46e4-89dc-a52731fea588 | 2026-05-05 | Veterans Affairs Employee Charged with Fraudulently Obtaini… | Veterans Affairs Employee | MA | PAYCHECK PROTECTION PROGRAM, PPP | 250000.0 | $250,000 | yes | wire fraud, false claims | charged | no | no identity-theft charge or third-party-identity language | medium | yes | Baez submitted two applications seeking Paycheck Protection… | https://www.justice.gov/usao-ma/pr/veterans-affairs-employe… | parsed | ||||
| 883d5034-e1e3-4f77-ac07-19388dfe10a3 | 2026-05-05 | Chicago Police Sergeant Charged in Federal Court With Fraud… | Chicago Police Sergeant | NDIL | PAYCHECK PROTECTION PROGRAM, PPP | 41662.0 | $41,662 | yes | false statements, wire fraud | charged | no | no identity-theft charge or third-party-identity language | medium | https://www.justice.gov/usao-ndil/pr/chicago-police-sergean… | parsed |
First 10 of 1,977 rows · cells longer than 60 characters are cut with an ellipsis · shown as text, unformatted.
Files
| File | Rows | Size | SHA-256 | |
|---|---|---|---|---|
ppp-eidl-identity-theft-classification.csvCSV, UTF-8, header row | 1,977 | 1,047,829 B | 4e7f846e17377d6d049caed33fd4c48f182875525083eec050e58a1630925bf1 | Download ↓ |
dictionary.csvdata dictionary, 21 rows | 21 | — | built with this page | Download ↓ |
Columns
| Column | Type, measured | Blank | Distinct | Range / values | Meaning |
|---|---|---|---|---|---|
uuid | text | 0 | 1,977 | The DOJ identifier; joins to the press-release index. | |
date | date (ISO) | 0 | 1,001 | 2020-05-13 – 2026-05-12 | Release date. |
title | text | 0 | 1,930 | Headline. | |
defendant | text | 311 | 1,069 | Defendant as extracted from the headline. | |
district | text | 191 | 88 | U.S. Attorney's Office district code. | |
programs | text | 1 | 103 | The index's programs for the same uuid. | |
dollar_amount | number | 98 | 904 | 1K – 832M · median $1M | The index's dollar_amount for the same uuid: the sentencing table's case amount when fact_source is canonical, otherwise the largest dollar figure in the release. |
dollar_amount_fmt | text | 98 | 904 | The same, formatted. | |
dollar_amount_reliable | yes/no | 333 | 1 | yes (1,644) | yes when the figure is a case amount rather than a program total or an unrelated figure; every canonical row with an amount is yes. |
charges | text | 79 | 565 | The index's charges for the same uuid. | |
case_type | text | 270 | 10 | sentenced (625), charged (351), pleaded guilty (348), indicted (134), convicted (93), arrested (57) | Stage keyword from the headline. |
sentence | text | 1,337 | 105 | The index's sentence for the same uuid. | |
restitution_fmt | text | 1,785 | 178 | Restitution as extracted, if any. | |
identity_theft | yes/no | 0 | 2 | no (1,731), yes (246) | yes or no. |
classification_basis | text | 0 | 17 | Which rule decided it: the charge, a text pattern, a hand override, or the default. | |
id_theft_subtype_note | text | 1,950 | 16 | For yes rows, the kind of identity misuse (stolen, synthetic, deceased, family, employee, …). | |
confidence | text | 0 | 3 | medium (1,181), high (795), low (1) | high, medium, or low. |
self_inflation_evidence | yes/no | 1,518 | 1 | yes (459) | yes when the text shows the defendant inflating their own payroll, headcount, or revenue. |
supporting_excerpt | text | 1,472 | 485 | The passage the classification rests on. | |
has_full_text | yes/no | 1,316 | 1 | yes (661) | yes when the full press-release body was available to the classifier. |
url | text | 0 | 1,977 | The justice.gov URL. | |
fact_source | text | 0 | 2 | parsed (1,364), canonical (613) | Copied from the index: canonical (fact columns from the sentencing table) or parsed. |
Type, blanks, distinct count, and range are measured on the file. The same table as CSV: dictionary.csv.
How to use it
Specific to this file, measured on it: keys, joins, encodings, units, and the values that will trip a naive count.
uuidis the key (0 duplicates) and every row joins to the DOJ press-release index; the index has 902 more rows (advisories, roundups, and non-PPP/EIDL programs) that are not classified here.- Three yes/no columns use blank for no.
dollar_amount_reliable,self_inflation_evidence, andhas_full_textcontain onlyyesor nothing (333, 1518, and 1316 blanks). Onlyidentity_theftwritesnoexplicitly. confidencetracks how much text was available: medium 1,181, high 795, low 1. A medium-confidence no was decided on the summary alone.dollar_amountis in dollars and comes from the index: the sentencing table's case amount on the 613 canonical rows, the largest figure in the release on the others, where program totals and district tallies can still leak in. Data release v17 corrected three such figures on rows marked reliable, a $903 million district total and nationwide totals of $836 million and $569 million. Its maximum is now $832 million, the capital a PPP lender borrowed through the PPP Liquidity Facility, on a row not marked reliable. Data release v18 corrected 74 more of its figures, 71 of them on rows marked reliable, and removed the reliable mark from 28 rows that have no amount. Sum only rows wheredollar_amount_reliableisyes.- CRLF line endings, UTF-8 with non-ASCII characters, no byte-order mark;
supporting_excerptcells run to 280 characters.
Before you quote a row
- These are readings of press releases, not legal findings. A no at medium confidence usually means only the summary was available and nothing in it pointed to a third party.
- Multi-defendant releases get one row. Where co-defendants' conduct differed, the row reflects the release as a whole.
- Program totals leak into dollar amounts; use
dollar_amount_reliablebefore summing. - Classifications describe the evidence available in the project corpus and are not legal findings.
Cite this table. Name the source vintage; the numbers depend on it.
PPP/EIDL identity-theft classification, built from the DOJ press-release corpus (classification through 2026-07-10). Pandemic Darlings, release pandemic-darlings-data-v1-2026-07-10, https://pandemicdarlings.com/data/ppp-eidl-identity-theft-classification/, SHA-256 4e7f846e17377d6d049caed33fd4c48f182875525083eec050e58a1630925bf1