Evidence desk / Paycheck Protection Program
Reported jobs are not jobs saved
An April 2020 claim, a July 2020 approvals count, and later assessments answer different questions. The Delaware loan slice checks only its own arithmetic.
Public by July 6, 2020
What the public record counted
April 28, 2020: President Donald Trump said, “Our swift action supported or saved 30 million American jobs at least.” This was a claim in PJ1, dated remarks, not a measured counterfactual. The PJ2 initial rule, April 15 described payroll retention, borrower certifications and documentation, not measured effects. Its initial terms subsequently changed.
July 6, 2020 release; approvals through June 30: 4,885,388 net approved loans, $521,483,817,756 net approved dollars, 51,125,937 jobs borrowers reported on applications, and 5,461 lenders. These reflect cancellations through the report date, not disbursement or a count of unique jobs permanently preserved. PJ3, July release; PJ4, pages 2 and 4.
The report compared its jobs figure with 59,900,000 Census small-business employees; that benchmark is not a causal control group. PJ4, page 4.
Published January 14, 2021 and April 25, 2022
What later work questioned and estimated
January 14, 2021: The PJ5 SBA inspector-general report, Finding 2 reviewed loan-system data as of June 30, 2020. It found 191,003 loans, totaling approximately $11,000,000,000, without information in the required current-employees field; 148,019 loans, totaling approximately $6,700,000,000, without current-employees and jobs-retained information; and 845,778 loans without jobs-retained data. These missing-field categories can overlap and must not be added. These data-quality findings do not prove every loan fraudulent or count jobs lost.
Published April 25, 2022: PJ6, Autor and coauthors, working paper abstract estimated an additional 3,600,000 jobs due to PPP as of mid-May 2020 and 1,400,000 at the end of 2020, using payroll data and eligibility cutoffs. They are causal estimates from one study, not the original borrower-reported workforce count, an observed census of saved jobs, or unique jobs guaranteed permanently. Different estimates may use different methods.
Open the audit's consequential distinction
July's application totals and the later audit's field checks are not interchangeable. Inspect the report's explanation of which field was mandatory.
Read PJ5's January 14, 2021 source excerpt about required and retained fields
“SBA did not require borrowers to provide data on jobs retained and did not require lenders to input data for jobs retained at the time of loan application. As a result, SBA officials and national leaders do not have enough information to make informed decisions or determine to what extent the PPP met national program objectives. Additionally, SBA cannot accurately report jobs retained by PPP borrowers, which was a statistic used in public reporting on PPP loans.”
This changes the interpretation: a missing jobs-retained entry need not mean zero employees, and the national reported-jobs count cannot establish an employment effect. PJ5, Finding 2, page 10; local excerpt.
Original-vintage Delaware subset / public July 6, 2020
A subset check, not a national jobs denominator
The PJ7 Delaware mirror of the original loan file covers loans of $150,000 and under only. The source field is JobsRetained, distinct from any later renamed JobsReported field. Nonblank sums below are recorded entries, not verified saved jobs; blanks are not observed zeros. PJ7 local aggregation and sample.
JobsRetained field within Delaware only. Reproducible source aggregation.| Delaware loan size | Loans | Loan amounts | Field present | Blank | Explicit zero | Nonblank sum |
|---|---|---|---|---|---|---|
| $50,000 and under | 7,643 | $148,773,586.78 | 6,693 | 950 | 1,439 | 20,706 |
| over $50,000 to $100,000 | 1,926 | $138,432,380.53 | 1,704 | 222 | 244 | 15,243 |
| over $100,000 to $150,000 | 862 | $104,952,525.95 | 756 | 106 | 79 | 11,465 |
| Delaware subset total | 10,431 | $392,158,493.26 | 9,153 | 1,278 | 1,762 | 47,414 |
The three bracket boundaries are $50,000 and under; over $50,000 to $100,000; over $100,000 to $150,000. Original CSV header and a documented 20-row spot-check sample appear in PJ7 local excerpt. These Delaware counts are not scaled to the nation.
The fixed-seed spot-check sample contains 20 Delaware rows totaling $1,203,772.60 in loan amounts. Of their original JobsRetained fields, 17 are present, 3 are blank and 3 are explicit zero; the nonblank sum is 148. This sample checks the excerpt, not a national estimate. PJ7 sample rows and method.
What the comparisons can and cannot say
The April 28 assertion, the June 30 approvals snapshot released July 6, the audit published January 14, 2021, and the research published April 25, 2022 are differently timed and defined. None is a directly comparable sequence of unique jobs saved. The research estimates an employment effect; the earlier jobs field records borrower reports; the audit checks completeness. The later results were not available to the July 2020 public.
Check the parsed figures and source IDs, national source excerpt, dated claim and rule excerpts, later excerpts, 2020 source ledger and later source ledger. Links to published records identify provenance; all evidence above remains readable offline.