Derived table · Paycheck Protection Program
PPP small loans: who got them, when, and what the fee rules paid for making them
How the smallest PPP loans got out, month by month: 151 cells of approval month, loan-size band and draw, each with the loans approved, the dollars behind them, the median loan, and the lender fee the published schedule would pay for making them.
Download CSV 24 KBData dictionary CSVAt a glancePreviewColumnsHow to useBefore you quote a row
At a glance
Measured on the served file on 2026-09-21; the checksum above is the file these numbers describe.
month + draw + band_index · uniqueWhat it is
One row is one approval month, one loan-size band and one draw. The eight bands are cut at $5,000, $10,000, $20,833.33, $50,000, $150,000, $350,000 and $2,000,000; every cut but the $10,000 one is a boundary in an SBA rule, in a fee schedule, or in SBA's own file layout, and the $10,000 cut is analytical and labeled as such in the companion file.
Three companion files are cut from the same universe and reconcile to it: small-loan-lenders.csv ranks the 50 largest originators of small loans under each of two definitions of small, fee-incentive.csv prices every band under both fee schedules, and geography.csv gives counts and shares by borrower state.
The unit is a loan, not a business, a borrower or a job. One business holding a first-draw and a second-draw loan is two loans here.
How it was built
The universe is every loan in SBA's release of 30 September 2024 — 11,468,210 rows across 13 files — less the four whose undisbursed amount is at or above their approval amount. That leaves 11,468,206 loans and $792,601,069,043.55 approved, dated 3 April 2020 to 19 July 2021. Within it, 8,984,989 loans were $50,000 or less and 7,306,941 were $20,833.33 or less, the Schedule C maximum.
Nothing is hand-entered. The 13 raw files are read with Python's CSV reader into one compact row per loan, then aggregated. Money is carried as integer cents and fees as integer ten-thousandths of a dollar, which is exact for the 50, 5, 3 and 1 percent rates and the flat $2,500, so the totals are not floating-point estimates.
Every figure on this page was recomputed from the loan file when the page was built: the universe, both small-loan thresholds, all 151 cells, the eight fee-incentive bands and all 60 state rows were re-derived and compared to the table, with 0 mismatches.
Preview
| month | draw | fee_regime | band_index | band_label | band_lower_usd | band_upper_usd | loans | dollars_approved | median_loan_usd | share_of_month_loans_pct | share_of_month_dollars_pct | fee_modelled_usd | fee_pct_of_principal |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2020-04 | First Draw | 2020 CARES | 1 | $1 – $5,000 | 0.01 | 5000.00 | 178175 | 568591582.64 | 3300.00 | 6.151867 | 0.134525 | 28429579.1320 | 5.000000 |
| 2020-04 | First Draw | 2020 CARES | 2 | $5,000.01 – $10,000 | 5000.01 | 10000.00 | 274454 | 2082937540.03 | 7517.64 | 9.476103 | 0.492809 | 104146877.0015 | 5.000000 |
| 2020-04 | First Draw | 2020 CARES | 3 | $10,000.01 – $20,833.33 | 10000.01 | 20833.33 | 549310 | 8732896608.23 | 15773.00 | 18.966086 | 2.066146 | 436644830.4115 | 5.000000 |
| 2020-04 | First Draw | 2020 CARES | 4 | $20,833.34 – $50,000 | 20833.34 | 50000.00 | 652682 | 21902671746.52 | 32516.31 | 22.535222 | 5.182029 | 1095133587.3260 | 5.000000 |
| 2020-04 | First Draw | 2020 CARES | 5 | $50,000.01 – $150,000 | 50000.01 | 150000.00 | 695110 | 60912372523.38 | 81995.00 | 24.000138 | 14.411469 | 3045618626.1690 | 5.000000 |
| 2020-04 | First Draw | 2020 CARES | 6 | $150,000.01 – $350,000 | 150000.01 | 350000.00 | 300566 | 67862274782.47 | 214400.00 | 10.377675 | 16.055771 | 3393113739.1235 | 5.000000 |
| 2020-04 | First Draw | 2020 CARES | 7 | $350,000.01 – $1,999,999.99 | 350000.01 | 1999999.99 | 219242 | 164213048592.98 | 608953.50 | 7.569792 | 38.851734 | 4926391457.7894 | 3.000000 |
| 2020-04 | First Draw | 2020 CARES | 8 | $2,000,000 and above | 2000000.00 | 26732 | 96390644491.38 | 2999325.00 | 0.922979 | 22.805396 | 963906444.9138 | 1.000000 | |
| 2020-04 | Second Draw | 2020 CARES | 4 | $20,833.34 – $50,000 | 20833.34 | 50000.00 | 3 | 97145.00 | 32842.00 | 0.000104 | 0.000023 | 4857.2500 | 5.000000 |
| 2020-04 | Second Draw | 2020 CARES | 7 | $350,000.01 – $1,999,999.99 | 350000.01 | 1999999.99 | 1 | 410655.00 | 410655.00 | 0.000035 | 0.000097 | 12319.6500 | 3.000000 |
First 10 of 151 rows · cells longer than 60 characters are cut with an ellipsis · shown as text, unformatted.
Files
| File | Rows | Size | SHA-256 | |
|---|---|---|---|---|
ppp-small-loan-access.csvCSV, UTF-8, header row | 151 | 24,611 B | f165f1ae1bd4f2e57a5f66fb53e519c5947bb2275e8d87dc878e15a5808694cb | Download ↓ |
dictionary.csvdata dictionary, 14 rows | 14 | — | built with this page | Download ↓ |
small-loan-lenders.csvthe 50 largest originators of small loans, under each of two definitions of small | 100 | 17,228 B | 112ae3b98706034247e0b94af5c78459cee6df0d321fec5d8bdb9cead090f183 | Download ↓ |
fee-incentive.csvevery band priced under both fee schedules, and loans of $50,000 or less by month | 19 | 4,286 B | 58b3dd7f2543fc26d6f27e6b2505478574ff2b485cb19199d15ba3fad8809aec | Download ↓ |
geography.csvloans of $50,000 or less by borrower state, with the state's whole book beside it | 60 | 16,273 B | 477a687e89d067fc4c4837ecaba9edc5548a61fe2625dec798c99f1c573846cb | Download ↓ |
README.mdwhat the table is, what it does not measure, and how every number can be checked | — | 5,385 B | 7ab7ec3a8ae94d770e15f2ff8e4adc87c83f4862660c7e3e24c028f183f5fa66 | Download ↓ |
SHA256SUMSSHA-256 of every other file on this page | — | 0 B | — | Download ↓ |
Columns
| Column | Type, measured | Blank | Distinct | Range / values | Meaning |
|---|---|---|---|---|---|
month | text | 0 | 13 | Calendar month of DateApproved, YYYY-MM. | |
draw | text | 0 | 2 | First Draw (90), Second Draw (61) | First Draw or Second Draw. Second Draw is ProcessingMethod = PPS in SBA's file. |
fee_regime | text | 0 | 3 | 2020 CARES (52), 2021 Economic Aid Act, first draw (50), 2021 Economic Aid Act, second draw (49) | Which published fee schedule prices the loans in this row: the 2020 CARES schedule, the 2021 Economic Aid Act first-draw schedule, or the 2021 second-draw schedule. |
band_index | integer | 0 | 8 | 1 – 8 · median 4 | Loan-size band, 1 (smallest) to 8 (largest). |
band_label | text | 0 | 8 | $10,000.01 – $20,833.33 (20), $20,833.34 – $50,000 (20), $350,000.01 – $1,999,999.99 (20), $1 – $5,000 (19), $50,000.01 – $150,000 (19), $5,000.01 – $10,000 (18) | The band as a dollar range. Every cut is taken from an SBA rule or a fee-schedule boundary except the $10,000 cut, which is analytical and says so in fee-incentive.csv. |
band_lower_usd | number | 0 | 8 | 0.01 – 2M · median 21K | Lowest loan amount in the band, in dollars. |
band_upper_usd | number | 17 | 7 | 5K – 2M · median 50K | Highest loan amount in the band, in dollars; blank on band 8, which is open-ended at $2,000,000 and above. |
loans | integer | 0 | 138 | 1 – 830K · median 17K | Loans approved in this month, draw and band, after the universe filter. |
dollars_approved | number | 0 | 151 | 200 – 164.2B · sum $792.6B · median $634.3M | Sum of CurrentApprovalAmount for those loans, summed in integer cents. |
median_loan_usd | number | 0 | 137 | 200 – 5.9M · sum $72M · median $33K | Median CurrentApprovalAmount in the cell. Where the cell holds an even number of loans this is the mean of the two middle loans, rounded to the nearest cent with halves going up. Every cell was recomputed in whole cents from the raw SBA files for this release. |
share_of_month_loans_pct | number | 0 | 144 | 0.00 – 100 · median 4.26 | The cell's loans as a percentage of every loan approved that month, both draws and all bands. |
share_of_month_dollars_pct | number | 0 | 151 | 0.00 – 100 · median 3.19 | The cell's dollars as a percentage of every dollar approved that month. |
fee_modelled_usd | number | 0 | 151 | 10 – 4.9B · sum $38.2B · median $47.4M | Lender processing fee for the cell, modeled from the schedule that governed those loans. Not a payment record. |
fee_pct_of_principal | number | 0 | 40 | 1 – 50 · median 5 | fee_modelled_usd as a percentage of dollars_approved. |
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.
- Fees are modeled, not paid. Every fee column applies a published schedule to a loan amount. SBA has never published per-loan or per-lender fee payments. Clawbacks, netting and fees on loans later cancelled or charged off are outside the model, which does not know whether a lender was paid at all.
- Loan size is not borrower size. A $15,000 loan means a small payroll base in the month used to compute it. It establishes nothing about the borrower's revenue, employment, sector or survival.
- Nothing here is a cause. The 2021 schedule paid far more per small loan than the 2020 schedule, and far more small loans were made in 2021. Both are measurements. The eligibility formula, the Schedule C gross-income change, second draws, set-asides and lender entry all changed at the same time.
- A platform network row is the partner lender's whole book. Where
platform_networkis filled in the lenders file, it names a lender that a held document identifies as that platform's partner. The loans and fees are the lender's, for its whole book of that size. - 17 mis-coded loans. SBA's file marks 17 loans as second-draw but dates them in 2020. Following the archive's standing treatment they are counted by approval date with the 2020 round and priced under the 2020 schedule.
- CRLF line endings, UTF-8, 24,611 bytes.
Before you quote a row
- Geography has no denominator. The state file gives counts and shares, not loans per 1,000 establishments: no 2019 establishment file is held, and a raw state count tracks state size before anything else.
- Lender types are blank. SBA's file has no lender-type field and the archive holds no SBA lender list. Calling a lender a fintech because of its name is not a measurement.
- Demographics are absent by choice. SBA's race, ethnicity, gender and veteran fields are largely unanswered in this release, so no subgroup comparison is published here.
- The same fee model, cut three other ways. PPP fee schedules by vintage gives it at three rows, the lender ranking ranks whole books, and the fee counterfactual prices one loan at a time.
PPP small loans: who got them, when, and what the fee rules paid for making them, compiled from SBA's loan-level release of 30 September 2024. Pandemic Darlings, https://pandemicdarlings.com/data/ppp-small-loan-access/, SHA-256 f165f1ae1bd4f2e57a5f66fb53e519c5947bb2275e8d87dc878e15a5808694cb