Editorial ruling / Paycheck Protection Program
Count the reported jobs. Do not call them saved.
An April assertion about jobs saved cannot be verified by a June approvals snapshot. The narrower, publishable fact is the number of jobs borrowers reported on applications.
Publish: By June 30, PPP had 4,885,388 net approved loans and borrowers reported 51,125,937 jobs on applications. Neither is a measured count of jobs saved. PJ4, pages 2 and 4, published July 6.
The claim / April 28, 2020
“Our swift action supported or saved 30 million American jobs at least. And last week, Congress answered our call to replenish the program, and I was honored to sign an additional $320 billion for American workers into law.”
President Donald Trump, East Room remarks, April 28, 2020 (PJ1). Exact record quotation.
How many of those jobs would have disappeared without PPP? Applications and approvals do not supply that counterfactual.
Two dates, two kinds of evidence
The remarks asserted at least 30 million jobs supported or saved. The April 15 initial rule stated a purpose of retaining workers and required borrower certifications and subsequent documentation. Those provisions describe intent and requirements, not a measured employment outcome. The original eight-week period and 75% payroll rule were later changed, so they are not described here as July requirements. PJ2, 85 FR 20812, 20814 to 20815.
The July 6 release described disclosure of the loan data. The official report covers approvals through June 30, both rounds and cancellations. $521,483,817,756 is net approved dollars, not verified disbursements or wages paid. PJ3, July 6; PJ4, pages 2 and 4.
The April first-round language and June 30 two-round net approvals do not have the same date, denominator or cancellation status. July evidence cannot establish what had already been saved in April.
National reconciliation, not an employment experiment
4,885,388
Sum of seven lender asset buckets equals the official national loan count.
51,125,937
Sum equals the national jobs figure, which is borrower-reported on applications. PJ4, pages 4 and 13.
That is 10.47 reported jobs per net approved loan, a ratio of two reported national totals, not an effect per loan. Against PJ4’s Census benchmark of 59,900,000 small-business employees, 85.35% is the arithmetic ratio. PJ4 prints “as much as 84%”; that figure does not follow from the printed 51,125,937 and 59.9 million without another denominator or adjustment. The July 6 PJ3 description “over 80 percent” is compatible with this ratio. Neither percentage measures jobs created or retained because of the program.
Inspect the seven lender asset-size rows / PJ4 page 13
These are lender assets, not the size of a borrower's loan. Dollars in this table are report-rounded lender bucket values; their sum is $521,600,000,000, not the precise national amount.
| Lender assets | Loans | Jobs reported | Lenders | Net approvals, rounded |
|---|---|---|---|---|
| $100,000,000 and under | 148,167 | 1,253,003 | 1,117 | $11,500,000,000 |
| over $100,000,000 to $500,000,000 | 522,037 | 5,102,159 | 2,623 | $43,800,000,000 |
| over $500,000,000 to $1,000,000,000 | 685,532 | 4,944,364 | 761 | $45,400,000,000 |
| over $1,000,000,000 to $5,000,000,000 | 890,007 | 9,698,005 | 733 | $95,200,000,000 |
| over $5,000,000,000 to $10,000,000,000 | 259,841 | 3,297,529 | 104 | $35,200,000,000 |
| over $10,000,000,000 to $50,000,000,000 | 739,912 | 9,595,633 | 89 | $100,700,000,000 |
| over $50,000,000,000 | 1,639,892 | 17,235,244 | 34 | $189,800,000,000 |
Lender count sums to 5,461, matching the national 5,461 in PJ4.
Inspect the separate national loan-size distribution / PJ4 page 3
Its eight loan-count rows sum to 4,885,388; their exact net approved dollars sum to $521,483,817,756. PJ4 gives no jobs-by-loan-size column. Printed one-decimal row percentages follow from the national counts and dollars when rounded individually; their displayed sums need not equal 100% after rounding. PJ4, pages 2 to 4.
| Loan size | Loans | Net approved dollars | Count share | Dollar share |
|---|---|---|---|---|
| $50,000 and under | 3,262,529 | $58,652,110,621 | 66.8% | 11.2% |
| over $50,000 to $100,000 | 673,563 | $47,963,195,310 | 13.8% | 9.2% |
| over $100,000 to $150,000 | 291,019 | $35,626,300,937 | 6.0% | 6.8% |
| over $150,000 to $350,000 | 376,113 | $84,452,629,388 | 7.7% | 16.2% |
| over $350,000 to $1,000,000 | 199,456 | $113,442,814,223 | 4.1% | 21.8% |
| over $1,000,000 to $2,000,000 | 53,030 | $73,522,278,271 | 1.1% | 14.1% |
| over $2,000,000 to $5,000,000 | 24,838 | $73,841,502,099 | 0.5% | 14.2% |
| over $5,000,000 | 4,840 | $33,982,986,907 | 0.1% | 6.5% |
What the smaller loans actually show
The PJ7 original-vintage Delaware file (third-party immutable mirror) covers loans of $150,000 and under, not the national June 30 snapshot. Its field is called JobsRetained. These are recorded fields, not verified jobs preserved. The plot shows only sums of nonblank entries, with a zero baseline. Blank entries are missing, not observed zero.
| Delaware loan size | Loans | Loan amounts | Field present | Field blank | Explicit zero | Nonblank jobs 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 subtotal: 10,431 rows; $392,158,493.26 loan amounts; 9,153 present jobs fields, 1,278 blank and 1,762 explicit zeros; 47,414 nonblank jobs-field sum. No extrapolation to the United States. PJ7 excerpt.
Inspect the record / choose a group
Choose a Delaware loan-size group to inspect its reported loan count, present fields and blank fields. The entire comparison remains in the table above without JavaScript.
Sample method and field limits
Twenty rows were selected by a fixed-seed random sample of all 10,431 Delaware data rows, sorted by original file position; the documented rows, exact sampling expression, source header and hashes appear in PJ7 excerpt and machine-readable reconciliation. The source header uses JobsRetained, not JobsReported. The original CSV cannot be rehashed offline from this packet: the excerpt supplies its claimed hash, while the excerpt's 20-row sample-block hash is locally recomputable. Its documented loan approvals run April 4 to June 30. No borrower is identified here.
What would establish an effect?
The 51,125,937 numerator is jobs listed on borrower applications; the 59,900,000 denominator is a Census small-business employment benchmark, not a matched contemporaneous control population. For the reported-jobs-per-loan ratio, the denominator is net approved loans, not unique workers or disbursed loans. A causal claim needs comparable pre- and post-program employment, actual disbursement and payroll use, overlap among counted workers, and a credible estimate of employment without PPP. None is observed in this packet. A discrepancy in a percentage is not evidence that aid had no effect.
Dated source marks
- PJ1, White House remarks, April 28, 2020.
- PJ2, Federal Register initial rule, April 15, 2020.
- PJ3, Treasury release, July 6, 2020.
- PJ4, SBA and Treasury report, approvals through June 30, released July 6, 2020.
- PJ7, original-vintage Delaware CSV, third-party mirror committed July 6, 2020; locally supplied source excerpt.