The Lending Scorecard

The scorecard table is the franchise figure — same columns, same denominator convention, same layout in every edition. FY2025 sits in the highlighted last row: 77,805 7(a) approvals, $37.22B gross, median loan size $167,900 (up 12% YoY against roughly 30% cumulative inflation over the full panel). The $37.22B number is the second-largest 7(a) year on record, and the only non-PPP year above $37B. 504 ran a parallel line — 6,759 approvals at $1.15M average, 71,863 jobs supported, up from 5,993 / $1.11M in FY24.

The rhyme to look for year to year is count versus dollars. A year where counts rise faster than dollars is a small-loan year; a year where dollars rise faster is a big-loan year. FY21 was the biggest dollar distortion in the panel. FY25 is the cleanest recent baseline.

FY2025 closed hot — 814 approvals on the final day and 1,404 across the two-day Sep 29–30 sprint as lenders cleared end-of-fiscal-year pipelines. Then the agency went dark. From Oct 1 through Nov 12, 2025 — 43 consecutive calendar days — SBA approved zero 7(a) or 504 loans. No prior stretch in the 10-year public lending record comes close. When operations resumed on Nov 13, the backlog released in a single session: 1,116 7(a) approvals versus an FY24 Thursday median of 279. The week of Nov 13–20 moved $1.84B in 7(a) volume, 1.98× the same week in FY24 ($927M).

Linear scale, not log. The 4× magnitude is the story; compressing it defeats the point. The week after Nov 13 wasn’t a steady return-to-normal glide. Nov 14 cleared another 915 approvals; Nov 15-16 fell to 101 and 20 (weekend); the rest of the month ran at or above typical pace. The backlog went out the door on the first two days.

The middle of the 7(a) market grew. The under-$50K micro end shrank. In FY2015 the under-$50K bucket was 28% of 7(a) count (17,447 of 62,854); in FY2025 it was 15.6% (12,120 of 77,805), even as the overall market added nearly 15,000 loans. The $50K–$350K mid-band ran the other way — up from 45.4% of count in FY15 to 50.7% in FY25, an absolute gain of nearly 41,000 loans. The $350K–$1M band went from 16.2% share to 21.6%. Above $1M crossed 12% share for the first time. Median loan size is up 43% nominally (FY15 $117K → FY25 $167,900) against roughly 30% cumulative inflation, so the real median is rising too, though not by as much as the headline suggests.

The FY21 spike on the largest bands is PPP-era policy distortion — the $5M guarantee ceiling and temporary-increase interactions. By FY23 the panel had reset. FY24 and FY25 show the underlying drift: SBA 7(a) is becoming less of a microloan program and more of a working-capital- and-acquisition platform for borrowers in the $150K-$1M band. That matches Chart E’s change-of-ownership surge.

Herfindahl–Hirschman Index on 7(a) funded volume climbed from 112 at the FY21 trough to 181 in FY25 — an 8-year high, up 62% in four years, and still well inside the DOJ “unconcentrated” zone (< 1,500). Active lender count fell from 1,451 to 1,308 over the same span, a 10% decline. The headline number travels, but the more useful fact is in the wedge chart below: the top-10 line for loan count (44.7% in FY25) sits well above the top-10 line for volume (32.0%), and the gap has been widening since FY22. Nearly half of every SBA 7(a) borrower in FY25 walked into one of ten lenders; only a third of the dollars went to those same ten names. The top-10 cohort specializes in smaller-dollar Express loans, so the borrower experience is materially more concentrated than the dollar-flow headline reveals.

Three of the top five 7(a) lenders are writing completely different deals. Live Oak Banking Company writes 2,270 loans at a $1.25M average; The Huntington National Bank writes 6,940 loans at $299K; Newtek Bank NA writes 4,815 at $421K; Northeast Bank writes 7,785 at $168K. Same league table, four strategies. Huntington is a high-count franchise / search-fund-scale platform. Live Oak is a verticalized large-deal specialist. Northeast is the Express / small-loan machine. The FY25 rank-by-volume list stays clustered at the top — Live Oak, Huntington, Newtek, Northeast, Readycap — but the underlying unit economics diverge more every year.

Even as concentration rose, the Fed cut Prime 91 bps in FY25 and lenders passed the cut through: vol-weighted spread held flat at roughly P+1.99%, median unchanged at P+2.75%. The system-wide rate mechanism worked. The real pricing problem is where on the dispersion curve a borrower lands — moving up one quintile of the lender spread distribution costs more than the entire FY25 Fed cut. That’s Section 4’s story, with the 514-bp all-lender range, the size-inversion (widest dispersion on the smallest loans), and the Newtek flat-ceiling finding.

Change-of-ownership 7(a) lending hit $8.80B in FY25 across 7,515 loans — both all-time records. Dollar volume up 31% in a single year. Share of 7(a) count rose from 8.63% in FY24 to 9.66% in FY25, reversing a three-year slide from the FY21 peak of 12.79%. This is the acquisition-entrepreneur cohort showing up in federal data at scale: searchers, self-funded SMB buyers, holdco operators. SBA 7(a) handles $1M+ acquisitions at 10% down with up to a 25-year amortization on real-estate-heavy deals. The FY23 trough (8.80% share) looks, in hindsight, like the bottom.

Cross-reference: Section 3 carries the podcast-cohort match study — of 990 acquisition deals publicly discussed on searcher podcasts, 2% match SBA acquisition records at publication-grade confidence, and when they do match the median loan (~$1.0M) is statistically indistinguishable from the SBA-wide acquisition median ($1.18M). The macro pattern in Chart E and the sampled-deal distribution in Section 3 tell the same story from different angles.