Section 05

The Borrower Playbook

In FY2025, 1-in-5 approved SBA loans never closed. In FY2018 it was 1-in-10. More deals reached the approval line. More of them died there.

Report-level thesis

18.87%FY2025 fall-outvs 10.46% in FY2018

1-in-5 SBA loans approved in FY2025 never closed. In FY2018 it was 1-in-10. The FY2018-FY2023 pooled baseline sits at 11.85% on n=353,522. FY2025 is a +8.41pp step-change from FY2018, and a +7.02pp jump from the multi-year baseline. One business owner handed a loan approval in FY2018 was twice as likely to actually close it as one handed an approval in FY2025.

(combined, FY2018 vs FY2025, n=150,445, all SBA (7(a) + 504) approvals; fall-out = loan_status IN ('CANCLD','NOT FUNDED'))

Thesis

Approval is cheap. Closing is the bottleneck.

Underwriting didn’t tighten at the approval line. SBA rubber-stamped more deals in FY2025 than in any non-PPP year on record. The six weeks between approval and close — the window a borrower treats as paperwork — became the most expensive part of the process. Columbia Bank closes 99 of 100 approvals. Northeast Bank closes 47. Bank Five Nine, the slowest major 504 first-lien partner, has a post-approval kill rate of 58%. Colony Bank closes its median 7(a) in 3 days. Webster Bank takes 42. Same program. Same public record. Same fiscal year.

Sub-sections below: 5.1 close-rate variance by program, lender, state, industry, size band. 5.2 which deal structures close, which die in diligence. 5.3 the time-to-close leaderboard — matters when a seller contract has a 60-day fuse. 5.4 the Borrower Scorecard, a four-question filter that produces three named lenders to call tomorrow. 5.5 the outlook: prime at 6.75% and falling, SOP 50 10 8 reset the acquisition-equity regime Jun 1, 2025, and FY2026 will be the first full year underwritten under the tight rules.

§ 5.1

Who Gets Approved, And Who Actually Closes

Two programs, two failure modes. The 7(a) fall-out rate in FY2025 was 15.77%; the 504 rate was 54.53%. Those are not measuring the same thing. Across the entire 16-year public lending record, no 7(a) loan has ever been coded NOT FUNDED; all 7(a) fall-out is CANCLD, which is SBA’s flag for an approved 7(a) that simply did not disburse — regardless of which side blinked. 504 fall-out is 95%+ NOT FUNDED, which is SBA’s separate flag for an approved 504 that failed the third-party diligence stack: appraisal gaps, environmental reports, first-mortgage bank pulling out, or the debenture pool underwrite not clearing. A 7(a) fall-out tells you the deal or the borrower blinked. A 504 fall-out tells you the deal structure broke.

Program	Approved	CANCLD	NOT FUNDED	Fell out	Pct
7(a)	77,805	12,271	0	12,271	15.77%
504	6,759	167	3,519	3,686	54.53%

(combined, FY25, n=84,564, all SBA approvals; fall-out split by loan_status)

Fall-out by size band — 7(a) FY2024-FY2025

Size is a weaker predictor than the plan initially read, but the $150K-$350K band — where Express-style quick-approvals meet real underwriting friction — still runs the highest at 14.88%. Bands above $500K, where structured acquisition deals dominate, close most cleanly. The sub-$150K band runs 13.56% — Express product, fast in, fast out, with the bulk of rate-shock cancellations.

Size band	Approved	Fell out	Pct	Typical use
$0-$150K	76,338	10,348	13.56%	Express / working capital
$150K-$350K	26,755	3,980	14.88%	Small equipment / partial acq
$350K-$500K	14,474	1,739	12.01%	Owner-operator acquisitions
$500K-$1M	13,602	1,355	9.96%	Small-biz acquisitions
$1M-$2M	9,069	1,034	11.40%	Mid-market acquisitions
$2M-$5M	7,612	811	10.65%	Upper-mid acquisitions + CRE

(7(a), FY24-25, n=147,850, 7(a) approvals with populated size_band)

The 7(a) lender fall-out leaderboard — FY2024-FY2025

This is the chart that screenshots. The five cleanest closers at the 200-approval-floor level all funded above 96% of their approvals; the five dirtiest include two of the largest 7(a) lenders in the country. Columbia Bank at 1.48% vs. Northeast Bank at 52.56% is a 35.5x gap on the same government guarantee. JPMorgan Chase sits on the cleanest list at 3.32% despite running 4,455 approvals — scale is not the reason for the variance. Newtek Bank’s 33.80% fall-out on 8,622 FY24-FY25 approvals is the volume-weighted number that pulls the national benchmark above 13%.

Cleanest closers (fall-out %, lowest-first)

Lender	Approvals	Fell out
Columbia Bank	1,354	1.48%
Central Pacific Bank	201	2.49%
Five Star Bank	251	2.79%
SouthState Bank, National Association	760	3.03%
JPMorgan Chase Bank, National Association	4,455	3.32%

Dirtiest closers (fall-out %, highest-first)

Lender	Approvals	Fell out
Northeast Bank	10,338	52.56%
Newtek Bank, National Association	8,622	33.80%
Bank of America, National Association	2,094	18.67%
Midwest Regional Bank	209	18.66%
Stone Bank	241	17.43%

This isn’t an underwriting-quality comparison. Columbia (n=1,354) runs a narrower credit box and tighter pre-qualification. Northeast (n=10,338) runs a broad-funnel preferred-lender model that treats approval as a screen, not a commitment. Both patterns are defensible from the lender’s side. What the borrower needs to know: an approval at a broad-funnel lender is a pre-screen with a coin-flip outcome, not a commitment you can plan around. See Section 2 for the full 151-lender league table with rate, spread, and vintage charge-off columns.

(7(a), FY24-25 pooled, n=147,850, lender floor n≥200 approvals)

The 504 first-lien partner table

Every major first-lien partner on 504 deals runs a post-approval kill rate above 25%. Bank Five Nine sits at 58.27%; Celtic Bank at 53.33%; M&T at 48.48%. This is the program, not the lender. The three-party 504 structure — first-mortgage bank + CDC + SBA debenture — produces this pattern regardless of individual lender quality, and the majority of fall-out shows up as NOT FUNDED (diligence-phase kill) rather than CANCLD (borrower walk-away). A 504 borrower who plans for 90 days and a crisp close has misread the program.

First-lien lender	Approved	CANCLD	NOT FUNDED	Fell out	Pct
Bank Five Nine	127	0	74	74	58.27%
Celtic Bank Corporation	135	5	67	72	53.33%
M&T Bank Corporation	132	6	58	64	48.48%
JPMorgan Chase Bank, National Association	390	14	148	162	41.54%
Bank of America, National Association	441	14	165	179	40.59%
Glacier Bank	137	1	50	51	37.23%
Meadows Bank	102	0	37	37	36.27%
Harvest Commercial Capital, LLC	287	6	94	100	34.84%
CalPrivate Bank	121	2	39	41	33.88%
First-Citizens Bank & Trust Company	368	6	111	117	31.79%
Zions Bank, A Division of	120	2	33	35	29.17%
Wells Fargo Bank National Association	157	1	39	40	25.48%

(504, FY24-25 pooled, n=13,517, first-lien lender floor n≥100)

Geography — fall-out by state

The national benchmark across the 43 states clearing the 500-approval floor is 13.02%. Alabama runs +4.66pp above that; Idaho runs -5.24pp below it. A borrower on one end of that spread closes differently than a borrower on the other end of the same program in the same fiscal year. The pattern tracks lender mix: hot states concentrate in the Southeast (AL, LA, GA, FL, NJ — where high-volume national broad-funnel lenders dominate) and cool states concentrate in the Upper Midwest and Mountain West (ID, IA, NE, OH, MN, MI, OR — where regional banks with in-market underwriters dominate). An Alabama borrower shopping at Huntington closes very differently than an Alabama borrower shopping at Northeast.

Hot states (above national, sorted highest-first)

State	Approved	Pct	Adj (pp)
AL ⚠	1,131	17.68%	+4.66pp
LA ⚠	1,197	16.21%	+3.18pp
GA	4,405	15.96%	+2.93pp
FL	12,184	15.89%	+2.87pp
NJ	5,316	15.69%	+2.66pp
CT	1,896	15.40%	+2.38pp
NY	10,089	14.88%	+1.85pp
MD	2,705	14.86%	+1.84pp
NV	1,618	14.77%	+1.75pp
MS	699	14.74%	+1.71pp

Cool states (below national, sorted lowest-first)

State	Approved	Pct	Adj (pp)
ID ⚠	1,452	7.78%	-5.24pp
PR ⚠	1,188	8.59%	-4.44pp
IA ⚠	848	8.61%	-4.42pp
NE ⚠	697	8.90%	-4.13pp
OH ⚠	7,141	9.06%	-3.96pp
MN ⚠	3,184	9.11%	-3.92pp
MI ⚠	5,169	9.34%	-3.68pp
OR ⚠	2,289	9.65%	-3.37pp
NM ⚠	570	10.00%	-3.02pp
IN	2,551	10.07%	-2.95pp

(7(a), FY24-25 pooled, n=147,557, states n≥500) · national benchmark 13.02%, ⚠ exceeds ±3pp

Industry — 2-digit NAICS spread

Information (sector 51) tops the industry table at 16.12% fall-out; Finance & Insurance (sector 52) sits at the bottom at 11.16%. That is a ~4.96pp spread, which is the finding: industry alone is a weak predictor. Sectors that hinge on third-party diligence (Information = customer-concentration and IP-license review; Real Estate = appraisal and environmental; Wholesale = inventory audit; Pro Services = book-of-business verification) cluster at the top. Hard-asset and working-operator sectors (Construction, Accommodation/Food, Health Care) cluster near the bottom. The 35x lender gap and the ~5pp size-band spread both swamp this signal by an order of magnitude.

NAICS	Sector	Approved	Pct
51	Information	1,501	16.12%
53	Real Estate & Rental/Leasing	3,030	15.02%
42	Wholesale Trade	6,402	14.90%
54	Professional, Scientific & Technical Services	15,519	14.59%
48-49	Transportation & Warehousing	7,457	13.64%
31-33	Manufacturing	8,576	13.34%
61	Educational Services	2,431	13.25%
44-45	Retail Trade	16,727	12.91%
81	Other Services	14,978	12.84%
56	Admin & Support, Waste Mgmt	8,801	12.70%
11	Agriculture, Forestry, Fishing	1,280	12.66%
72	Accommodation & Food Services	17,759	12.50%
62	Health Care & Social Assistance	14,437	12.41%
23	Construction	20,403	12.14%
71	Arts, Entertainment & Recreation	5,603	11.44%
52	Finance & Insurance	2,384	11.16%

(7(a), FY24-25 pooled, n=147,331, 2-digit NAICS sector; floor n≥300)

§ 5.2

What Structures Close

Term length — the 7-to-10-year danger zone

The 7-to-10-year band is where 74% of 7(a) volume sits, and it runs 14.38% fall-out — the highest of any term band by a meaningful margin. Every other bucket sits at roughly 9% fall-out. The mechanism is goodwill-heavy acquisition financing (SBA caps goodwill; appraisals must justify the purchase price; QoE adjustments cut the seller’s EBITDA number by 10-40% on diligence review). Short-term loans (1-7 years, working capital and lines) carry the highest charge-off rate in the seasoned cohort at 0.35%, because borrowers reaching for short terms to shave interest cost end up with payment schedules that break in Year 3 when a quarter goes sideways.

Term	Approved	Fell out	Pct	Chg-off	Chg-off %
1-7yr	16,565	1,530	9.24%	58	0.35%
7-10yr	109,323	15,723	14.38%	233	0.21%
10-15yr	6,268	579	9.24%	0	0.00%
15-25yr	13,528	1,222	9.03%	0	0.00%
25yr+	2,166	213	9.83%	0	0.00%

(7(a), FY24-25, n=147,850, term_months > 0)

Business age — the acquisition-closes-cleanest finding

The counterintuitive finding: change-of-ownership deals (acquisitions) close cleanest at 7.70%, while established-business expansion loans fall out most at 15.53% — roughly 2x the acquisition rate. Borrowers and brokers assume acquisitions are the risky thing and existing-business expansions are the safe thing; the data says the opposite at the close line. The mechanism is structural: an acquisition comes with a signed purchase agreement, a seller-imposed deadline, a QoE report, and capital-stack terms that are already on paper. An expansion loan has none of that — the borrower can walk at any time, and often does when the full underwriting cuts the pre-approval’s amount or rate.

Business age	Approved	Fell out	Pct
Existing or more than 2 years old	88,263	13,709	15.53%
New Business or 2 years or less	23,508	2,640	11.23%
Startup, Loan Funds will Open Business	22,261	1,841	8.27%
Change of Ownership	13,559	1,044	7.70%

(7(a), FY24-25, n=147,591, business_age populated)

Interest rate — a timing artifact, mostly

The 10-11% rate band runs the highest fall-out at 16.59%, and the sub-9% bands run the lowest. The caveat: this is mostly a timing artifact. FY2024-FY2025 prime ran 7.5-8.5%, so most variable-rate 7(a) loans landed in the 10-11% band (Prime + 1.5 to 3). The higher fall-out there reflects the peak-rate cohort running into sticker shock at closing, not a causal rate effect. For cross-lender spread analysis, Section 4’s Chart 3 shows identical $2M HVAC-acquisition borrowers receiving spreads from P+0.75% to P+2.95% — a borrower who gets the prime-rate call wrong loses ~50bp over a variable period; a borrower who picks the wrong lender on spread loses 220bp for the full term.

Initial rate	Approved	Fell out	Pct
<8.00%	9,229	745	8.07%
8.00-8.99%	11,223	872	7.77%
9.00-9.99%	20,981	1,723	8.21%
10.00-10.99%	46,943	7,787	16.59%
11.00-11.99%	29,706	4,417	14.87%
12.00-12.99%	10,039	1,062	10.58%
13.00-13.99%	9,459	1,221	12.91%
14.00%+	10,264	1,440	14.03%

(7(a), FY24-25, n=147,844, non-null initial_interest_rate)

Collateral — a non-finding worth calling out

Collateralized loans fall out at 12.87% vs. 13.79% for uncollateralized. Less than 1pp gap. The "you need collateral to close" myth is not supported by the data. Collateral is a credit-quality and post-close default lever, not a close-rate lever — the SBA guarantee already covers the lender’s downside at the approval stage, so collateral shows up later, at charge-off recovery, not at funding.

Collateral	Approved	Fell out	Pct
Collateralized	121,700	15,662	12.87%
Uncollateralized	26,150	3,605	13.79%

§ 5.3

The Time-Is-Money Table

Median 7(a) close time in FY2025 was 21 days. Median 504 close was 119 days. At the lender level the 7(a) spread runs from 3 days (Colony Bank, SouthState, Eastern) to 62 days (Bank Five Nine), which is the difference between a seller’s 45-day contract surviving and not surviving. Webster Bank (42-day median on n=381) and M&T (41-day median on n=3,387) represent the large regional pattern where internal credit-committee review adds 4-6 weeks relative to the in-house underwriting of SouthState or Colony.

Approval → first disbursement, by program and FY

Program	FY	N funded	Mean days	Median days	P90 days
7(a)	2024	59,564	35.4	17	74
7(a)	2025	57,878	32	21	71
504	2024	4,115	212.3	147	455
504	2025	3,073	139.1	119	238

Fastest closers (median days, lowest-first)

Lender	Approvals	Median
Colony Bank	676	3 d
City National Bank of Florida	201	3 d
SouthState Bank, National Association	727	3 d
Eastern Bank	657	3 d
BayFirst National Bank	4,492	5 d
Cadence Bank	1,527	6 d
Celtic Bank Corporation	2,236	8 d

Slowest closers (median days, highest-first)

Lender	Approvals	Median
Bank Five Nine	251	62 d
Webster Bank National Association	381	42 d
Manufacturers and Traders Trust Company	3,387	41 d
Regions Bank	223	35 d
Wells Fargo Bank National Association	2,923	35 d
Wilmington Savings Fund Society FSB	201	34 d

The borrower action: for acquisitions, factor the lender’s median close-time into the seller’s-contract window with a P90 buffer. A seller granting 45 days plus a lender with a 42-day median is almost certainly a time-kill fall-out. The Section 2 league table carries close-time alongside rate, spread, and charge-off columns — the composite is where borrower cross-lender comparison actually lives.

(combined, FY24-25, n=117,442, funded loans with disbursement within 730 days)

§ 5.4

The Borrower Scorecard

Four questions, each a filter rather than a score. Q1 sets the deal type (which base fall-out rate applies). Q2 produces a size-matched lender pool. Q3 filters that pool by timeline (close-time cutoff removes lenders whose P90 would blow the seller deadline). Q4 ranks within that filter by state fit (adjust by the state-adjustment pp from 5.1; bias toward in-state underwriting where the state runs hot). The sequence matters; scores collapse too quickly to act on.

Q1 — What kind of deal?

Acquiring a business — base fall-out 7.70%, optimistic cohort. Go to Q2.
Refinancing or expanding existing — base fall-out 15.53%, budget 2x time. Go to Q2.
Starting from scratch — base fall-out 8.27%, expect heavy underwriting. Go to Q2.
Real-estate + building (504 candidate) — stop. 504 base fall-out 54.53%, median close 119 days. Read 5.1d before going to Q2.

Q2 — Deal size?

<$500K — 7(a) SBA Express territory. Speed beats pedigree. Pick by close-time leaderboard (5.3).
$500K-$1M — 7(a) standard. Mid-band fall-out at 9.96%, many lender options. Pick by close-rate leaderboard (5.1c).
$1M-$2M — 11.40% fall-out, acquisition-favored band. Pick from the specialist leaderboard below.
$2M-$5M — 10.65% fall-out, upper-mid acquisition + CRE. Pick 2-3 specialist backups, not one.
$5M+ — call a human. Public sample thins here.

Q3 — Timeline?

Seller deadline <60 days — need median close <14 days + fall-out <8%. Shortlist: Colony Bank, SouthState, Eastern, BayFirst, Celtic.
60-90 days — most 7(a) lenders work. Filter on fall-out, not speed.
90+ days or flexible — 504 is on the table; slower large regionals viable if the deposit relationship matters.

Q4 — State? (adjustment = state pct minus 13.02% national benchmark)

Adj ≥ +3pp (AL, LA flagged ⚠) — first-order risk factor. Bias strongly toward in-state underwriting; require a state-specific lender volume quote.
Adj +1 to +3pp (GA, FL, NJ, CT, NY, MD, NV, TN, VA, CA) — bias toward in-state underwriting. Add state adj to your cohort rate.
Adj -1 to +1pp — geography is not load-bearing. Pick on close-rate and lender fit.
Adj ≤ -3pp (ID, IA, NE, OH, MN, MI, OR flagged ⚠) — regional banks and community banks often outperform the national leaderboard at your size.

Q5 — Does this lender sell more than 90% of its variable-rate 7(a) book over the last three fiscal years?

If yes, the rate quoted to you is priced against the secondary market's yield demand, not the lender's cost of funds. That is not a reason to walk — the SBA guarantee is identical — but it changes what the lender can legitimately cite as cost-of-funds justification for a premium. Shop a hold-to-maturity peer for comparison. Section 6 — Who Flips Your Loan — names the top holders and flippers.
If no (flip rate ≤ 5%), the lender funds itself through deposits and prices to its own cost of capital. The premium dynamic does not apply to your quote.

The acquisition-specialist leaderboard

7(a), Change of Ownership only, floor n=100 approvals. U.S. Bank tops the table at 1.61% fall-out on 124 acquisition approvals. Huntington National Bank sits at 3.25% on 1,383 acquisitions — volume-weighted acquisition specialist, clearly underwrites acquisition deals at scale. Live Oak Banking Company, the other large acquisition specialist in the cohort, runs 8.59% on 1,374 acquisitions — still below the 12.99% program-wide average, but structurally 2.6x Huntington’s rate. Neither lender is wrong; different playbooks, different deal-book mixes.

Lender	Acq approvals	Fell out	Pct
U.S. Bank, National Association	124	2	1.61%
Old National Bank	207	6	2.90%
The Huntington National Bank	1,383	45	3.25%
Brookline Bank, a Division of Beacon Bank and Trust	162	7	4.32%
First Financial Bank	108	5	4.63%
First Internet Bank of Indiana	360	17	4.72%
T Bank, National Association	104	6	5.77%
GBank	244	15	6.15%
Readycap Lending, LLC	135	9	6.67%
Zions Bank, A Division of	116	8	6.90%
Celtic Bank Corporation	183	13	7.10%
Hanmi Bank	233	18	7.73%
Truliant FCU	142	11	7.75%
Byline Bank	261	22	8.43%
Live Oak Banking Company	1,374	118	8.59%

(7(a), FY24-25, n=13,559, business_age='Change of Ownership'; floor n≥100)

Six deal-killer patterns

Editor judgment applied to FY2024-FY2025 fall-out patterns and the regulatory regime change at SOP 50 10 8 (effective Jun 1, 2025). Each row links a specific structural pattern to its close-rate cost and a fix the borrower can apply before filing. These are the archetypes the two things will kill this deal summary line draws from.

Equity injection below 10%

What it is:: Buyer shows up with 5-8% cash down, expecting to structure the rest as seller financing or gift funds.
Regulatory anchor:: SOP 50 10 8 §A(3)(c) — minimum 10% equity required on change-of-ownership, owner-operator real estate, or business acquisition deals.
What changed:: Prior SOP permitted a broader interpretation of equity; current SOP hard-codes 10% with limited exceptions.
Borrower consequence:: Lender approves conditionally assuming equity at closing; borrower can't bridge the gap; loan moves to CANCLD.
Fix:: Verify cash-on-hand (including post-closing reserves) equals 12-15% of project cost before submitting. Budget for closing costs plus SBA guarantee fee.

Seller note structured as equity

What it is:: Seller holds a second-position note that pays sooner than 24 months or bears interest during the SBA loan term.
Regulatory anchor:: SOP 50 10 8 §A(3)(c) — seller notes counted toward equity must be on full standby (no principal or interest payments) for the first 24 months.
What changed:: Full-standby period is 24 months firm; prior SOP allowed shorter standby in some deal structures.
Borrower consequence:: Credit committee flags the note terms; the deal is approved then unwinds when seller refuses to amend.
Fix:: Draft seller notes as 24-month full-standby from day one. Negotiate seller's price expectation knowing the standby is non-negotiable.

Appraisal below purchase price

What it is:: Real estate or business appraisal comes in 5-15% below the signed purchase price.
Regulatory anchor:: SOP 50 10 8 §E — SBA will not guarantee above appraised value; shortfall must be covered by additional equity or seller price reduction.
What changed:: No change — but 2024-25 valuation softness surfaced this friction more often.
Borrower consequence:: Gap is $40-80K on a $1M deal; if seller won't reprice and buyer won't add equity, deal moves to NOT FUNDED.
Fix:: Include an appraisal contingency in the purchase agreement. Hold 5% of purchase price in reserve as 'valuation cushion' separate from your 10% equity.

Quality-of-earnings re-cut downward

What it is:: Seller-provided EBITDA is $450K; QoE report adjusts it to $290K after add-back scrub.
Regulatory anchor:: SOP 50 10 8 §E(2) — debt service coverage must be documented with third-party financial statements; lender bank discretion on QoE depth.
What changed:: More lenders are requiring independent QoE on acquisition deals above $1M (not mandated by SOP, but market practice post-2024 defaults).
Borrower consequence:: Revised DSCR falls below 1.25x; lender's credit committee requires either a larger equity check or a price concession.
Fix:: Hire your own QoE before LOI signing. Negotiate price based on scrubbed earnings, not seller's adjusted number.

Personal guarantee refusal (20%+ owners)

What it is:: One of multiple owners declines to personally guarantee the loan.
Regulatory anchor:: SOP 50 10 8 §B(2) — all owners holding 20%+ must personally guarantee; no waivers.
What changed:: Guarantee requirement was tightened under the current SOP; prior informal carve-outs eliminated.
Borrower consequence:: Deal stalls; one owner must either buy out the holdout or reduce their stake below 20%.
Fix:: Confirm all 20%+ owners' willingness to PG before submitting. For family-owned entities, map the cap table before loan application.

Borrower walk-away on rate or terms

What it is:: Between approval and closing, borrower finds a conventional product at 50-100bps lower or decides the deal economics no longer work.
Regulatory anchor:: None — this is borrower economics.
What changed:: Prime falling from 8.50% (Jul 2023) to 6.75% (Dec 2025) created a 6-month window where conventional alternatives became more competitive.
Borrower consequence:: CANCLD — clean exit, no regulatory issue, just a better offer elsewhere.
Fix:: Model your all-in cost including SBA guarantee fee vs. conventional before accepting approval. Decide pre-approval, not at closing.

§ 5.5

12-Month Outlook

Prime sits at 6.75% as of 2026-04-16. The Fed cut 25bp on Dec 10, 2025 and prime moved effective Dec 11. Total move since the 8.50% peak (Jul 27, 2023) is -175bp over 28 months; FY2025 alone delivered -75bp across three cuts. A 7(a) variable-rate loan approved at peak (Prime + 2.75 = 11.25%) vs. one approved today (9.50%) carries a ~$90/mo payment reduction per $100K borrowed on a 10-year amortization, or roughly $900/mo at $1M borrowed — enough to move a 1.18x DSCR into 1.28x on the same cash flow.

Prime rate inflection dates (last 7)

Effective date	Prime	Move	Direction
2025-12-11	6.75%	-25bp	cut
2025-10-30	7.00%	-25bp	cut
2025-09-17	7.25%	-25bp	cut
2024-12-19	7.50%	-25bp	cut
2024-11-08	7.75%	-25bp	cut
2024-09-19	8.00%	-50bp	cut
2023-07-27	8.50%	+25bp	hike

Claim 1 — the prime path is down, but spread is where the dollars hide. Fed dot plot (Dec 2025 SEP) implies 50bp additional cuts through end of FY2026, putting prime in the 6.25% range by Sep 2026 if realized. A variable-rate 7(a) filed today at Prime + 2.5% resets ~50bp lower over that cycle. That matters, but it is the small lever. The big lever is spread: Section 4 showed the same $2M HVAC-acquisition borrower receiving spreads from P+0.75% to P+2.95% at different lenders — 220bp of dispersion, worth $283K in lifetime interest on the same deal. A borrower who picks the wrong lender on spread loses that 220bp for the loan’s full term. A borrower who gets the rate call wrong loses ~50bp over the variable period. Rank the levers accordingly.

Claim 2 — FY2026 YTD is unreadable, but the FY2025 spike is not shutdown-contaminated.Two noise questions, two separate answers. FY2026 Q1 cannot be read as a reversion signal — the 43-day federal shutdown (Oct 1 - Nov 12, 2025) zeroed out approvals for six weeks, then produced a 1,116-approval clearance on Nov 13. Any FY2026 Q1 "trend" is backlog mechanics, not borrower-market signal; we will not know whether FY2025’s 18.87% fall-out is a peak-rate artifact or a new baseline until FY2026 Q2 data arrives in Apr-Jun 2026. Separately, the FY2025 spike itself is not a shutdown artifact — the sensitivity-gate Test B result (18.47% fall-out on pre-Jul 1, 2025 approvals vs. 18.87% full-year, 40bp delta) shows the spike held in the pre-shutdown cohort. Don’t bet your filing timeline on a reversion assumption, and don’t discount FY2025 as shutdown noise. Plan for FY2025-level close-rate risk and pick a lender whose historical fall-out is below 8%.

Claim 3 — SOP 50 10 8 flipped the acquisition-equity regime.Two SBA Standard Operating Procedures bracket the FY2025 fall-out spike; causation is inferred from timing, not proved. Aug 1, 2023 — SOP 50 10 7 took effect. Nov 15, 2023 — SOP 50 10 7.1 replaced it, removing specific equity-injection guidance in favor of a "do what you do" policy that let lenders apply their own non-SBA commercial standards. This was the loose regime. Jun 1, 2025 — SOP 50 10 8 took effect, reinstating pre-2021 standards: a firm 10% equity injection on change-of-ownership and startup transactions, seller notes capped at 50% of required equity and required to be on full standby for the entire SBA loan term (no principal, no interest), plus tightened personal-resources and character/citizenship review. Primary source: SBA Information Notice 5000-868665, Issuance of SOP 50 10 8 with Technical Updates. FY2025 was approved under the loose regime; FY2026 is the first full cohort underwritten under the tight one. Expect a hard 10% cash-equivalent equity requirement and seller notes on full standby for the loan term. The deal structure that cleared in 2024 is not the deal structure that clears in 2026.

Claim 4 — 504 becomes the rate play, with a caveat. With prime falling, 504’s fixed-rate debenture becomes relatively less attractive vs. a variable 7(a) that will reset down. For borrowers who value payment certainty, 504 is still the cleanest 25-year fixed-rate product available to small business. Only pursue it if your timeline can absorb a 120+ day close and a coin-flip probability on the way (5.1d). If the seller won’t wait, a 7(a) close in 21 days at a slightly higher rate is often the better answer.

§ 5.6 — Ask Riley

Questions Readers Have Actually Asked

First-person, brief. These are questions I get from borrowers and early readers of this report. The answers lean on what the data above shows, not on what I wish it showed.

Q. My lender just gave me an approval. How do I know if I should celebrate?

Check the lender’s FY2024-FY2025 fall-out rate before you sign anything. If it is under 8%, the approval is close to a commitment — you can plan your seller deadline around it. If it is 15%+, treat it as a pre-screen with serious downside risk; ask the loan officer directly "of your last 100 approvals my size, how many funded?" If they can’t answer, that is the answer. Columbia Bank clears 99 of 100; Northeast Bank clears 47. Both sent approval letters that looked identical.

Q. I’m buying a business for $1.8M. The seller gave me a 60-day close window. Is that realistic?

Realistic with the right lender, precarious with the wrong one. Median 7(a) close is 21 days; P90 is 71 days. If you’re working with Webster Bank (42-day median) or M&T (41-day median), your 60-day seller window is at coin-flip risk from the lender side alone, before counting the appraisal turn and the QoE review. With Colony Bank, SouthState, or Eastern (3-day median each), you have five-plus weeks of cushion. Pick the lender with the shorter tail and file two weeks before you think you need to.

Q. I keep hearing 504 has a lower rate than 7(a). Why wouldn’t I just go 504?

Because 504 has a 54% post-approval kill rate and a median 119-day close in FY2025. Those are not rounding errors. Every major 504 first-lien bank — Bank Five Nine, Celtic, M&T, JPMorgan Chase, Bank of America — runs above 25% NOT FUNDED, which is SBA’s diligence-phase-kill flag. The rate savings on 504 evaporate if the deal dies in appraisal, environmental, or CDC coordination six months in. 504 is the right product when you have real estate, a 150-plus-day timeline, and cash-flow tolerance for the holding-cost risk. It is a bad product when you treat it like a 7(a).

Q. Should I wait for rates to drop further before filing?

I don’t forecast rates, and neither does anyone who is honest. Prime went from 8.50% to 6.75% over 28 months; the Dec 2025 dot plot implies another 50bp of cuts, which may or may not happen. Here is what I tell people: if your deal is an acquisition with a seller deadline, you don’t have the option to wait, so pick the lender with the cleanest close record and file now. If your deal is an expansion on your own timeline, the bigger lever is the spread, not the rate path. Section 4 showed 220bp of lender-level spread dispersion on comparable deals. That’s worth more than the next 50bp of Fed cuts. Shop lender, not timing.

Methodology

Methodology & cross-section references

Data Provenance

Source: SBA 7(a) and 504 loan-level public releases, joined through LenderHawk’s lender canonicalization layer. All queries reproducible from public.loans, public.lenders, public.naics_clusters, and public.prime_rates in Supabase project xmaymzprormodlcsoqxs.

Fall-out definition: loan_status IN ('CANCLD','NOT FUNDED') at time of data pull. Excludes EXEMPT (active, too new to classify), COMMIT (in-flight), PIF (paid in full), and CHGOFF (funded, later defaulted). Program-split verified against 2010-2026 public data: 7(a) fall-out is 100% CANCLD; 504 fall-out is 95%+ NOT FUNDED.

Approval cohort: approval_fy = SBA fiscal year (Oct-Sep). Close time: first_disbursement_date - approval_date, funded loans only, capped at +730 days to exclude data anomalies.

Sample floors applied: 7(a) lender leaderboard n≥200; 504 first-lien n≥100; acquisition specialists n≥100; close-time leaderboard n≥200 funded; state adjustment n≥500; 2-digit NAICS n≥300.

Regulatory references: SBA Information Notice 5000-868665 — Issuance of SOP 50 10 8 with Technical Updates (effective Jun 1, 2025; Office of Capital Access).

Cross-section references: Section 1 (The Lending Scorecard) for FY2025 market shape and the shutdown anomaly; Section 2 (The Lender League Table) for full 151-lender cross-columns (rate, spread, charge-off); Section 3 (Money Flow) for state/industry $ flow context; and Section 4 (The Cost of Capital) for lender-level spread dispersion referenced throughout 5.2c and outlook Claim 1.

Window note on Northeast Bank and Newtek Bank: Section 2 reports each on an FY2025-only window (Northeast 61% on n=7,785; Newtek 38.9%); this section pools FY2024-FY2025 to smooth year-specific noise in the broader leaderboard (Northeast 52.56% on n=10,338; Newtek 33.80% on n=8,622). Both are correct under their denominators; the pooled window is applied consistently here.

How to read this

Fall-out, defined: Approvals where loan_status is CANCLD or NOT FUNDED at the time of data pull — the "approved but never disbursed" cohort. 7(a) fall-out is 100% CANCLD; 504 fall-out is 95%+ NOT FUNDED. The program-split is SBA’s categorization convention, not an analyst choice.
Sensitivity gate: The 1-in-5 headline ships only after three pre-render tests pass: trendline (is FY25 a step-change?), shutdown contamination (does excluding post-Jul 1, 2025 approvals change the answer?), and monthly cohort (is the spike a clean inflection or a single-month artifact?). All three passed at render.
Windows: Hero compares FY2025 (n=84,564) to FY2018 (n=65,881). Baseline for reproducibility is FY2018-FY2023 pooled (11.85% on n=353,522). Lender/state/industry leaderboards pool FY2024-FY2025 to stabilize samples.
Sample floors: 7(a) lender leaderboard: n≥200. 504 first-lien lender: n≥100. Close-time leaderboard: n≥200 funded. Acquisition specialists: n≥100. State adjustment table: n≥500. Industry: n≥300 at 2-digit NAICS.
What this section is not: Not a shame list. Close-rate variance has legitimate drivers: broad-funnel preferred-lender models treat approval as a screen, not a commitment. We report the delta and let the borrower weight it.

Full methodology →