Prepared by Michal Mohelsky, J.D., Practicing Affiliate of the Appraisal Institute, FMVA

MMCG Invest, LLC · San Francisco, CA

Published June 8, 2026 · Last updated June 8, 2026

The standing benchmark for project feasibility.

Feasibility Index - Q2 2026

1. Executive summary

The U.S. Small Business Administration sits on the largest public record of small-business credit performance in the country, and as of its March 31, 2026 disclosure that record reveals, for the first time in over a decade, the realized fate of nearly every loan. The MMCG Feasibility Index reads that record in full. This inaugural issue analyzes 1,036,077 loans: 919,732 in the 7(a) program and 116,345 in the 504 program, spanning approval years 2010 through the first quarter of 2026, representing roughly 401 billion dollars of 7(a) approvals and 97 billion dollars of 504 debentures on an estimated 228 billion dollars of total project cost (1)(2). Across the funded book, lenders and the agency have realized close to 6.0 billion dollars in charge-offs.

Five findings define the issue.

First, survival is governed by structure, price, and asset class far more than by the business concept. Once loans are old enough to have resolved, a 7(a) loan charges off at 6.44% and a 504 loan at 1.49%, and that gap holds in every industry sector without exception.
Second, the price of a loan is a near-perfect map of its risk. A 7(a) loan priced above 11% charges off roughly five times as often as one priced at or below 4%, and the relationship climbs in almost unbroken steps.
Third, the base rate that determines a feasibility conclusion lives at the industry, not the sector. Among the most active SBA industries the seasoned adverse-resolution rate ranges from under 1% for self-storage to above 20% for electronics retail, and the dispersion inside a single two-digit sector routinely exceeds the dispersion between sectors.
Fourth, SBA Express, which is half of the 7(a) program by count, does not raise default everywhere. It raises it sharply in capital-intensive businesses that normally rely on real-estate collateral, and it lowers it in labor-light service trades.
Fifth, the picture is forward-looking. Active distress on the still-open book is concentrated in the 2022 and 2023 origination cohorts, the loans made into the rate-shock window, which are the leading edge of the next wave of losses.

Each of these is drawn from the complete public population, not a sample, and each is presented below with the exhibit that supports it.

2. The dataset and the disclosure

For most of the program's modern history, the public SBA loan files disclosed approval terms but withheld outcomes. Workout states, the codes that record delinquency, deferment, liquidation, guaranty purchase, and charge-off, were treated as exempt from disclosure. The quarterly refresh dated March 31, 2026 changed that quietly: the loan-status field now populates the full set of resolution codes across the entire history of the file, retroactive to at least fiscal 2010, even though the published data dictionary still describes the old, narrower set (19). The practical effect is that realized credit performance, charge-offs, liquidations, guaranty purchases, delinquencies, and deferments, can now be measured by industry, lender, vintage, geography, and loan size across more than a million loans (1).

This issue treats the disclosure as what it is: a near-census of program performance, not a survey or a convenience sample. Its one material restriction is methodological, a seasoning adjustment that is the single most important analytical choice in the entire Index, and the analysis that follows applies it throughout.

3. The policy backdrop: a program under scrutiny

The disclosure does not arrive in a vacuum. It lands in the middle of the most consequential period of scrutiny the 7(a) program has faced in more than a decade, and that context is what gives a historical loss record its present-tense weight.

The proximate trigger was financial. The agency reported its first negative 7(a) program cash flow in over a decade in fiscal 2024, on the order of 397 million dollars, and identified deteriorating credit performance as the cause (7). Oversight sharpened the concern rather than easing it. In a March 2026 report the agency's Inspector General questioned costs on 73,302 loans totaling roughly 32 billion dollars (5), and for a sixth consecutive year issued a disclaimer of opinion on the agency's financial statements, meaning the accounts could not be certified to a clean audit standard (6).

The credit-standard response was specific. SOP 50 10 8, effective June 1, 2025, reaffirmed the long-standing requirement that a standard 7(a) loan above 350,000 dollars carry debt-service coverage of at least 1.15 to 1, and it lowered the ceiling for the streamlined 7(a) small-loan category to 350,000 dollars. The genuinely new floor came later: a March 2026 procedural notice set, for the first time, a numeric coverage minimum of 1.10 to 1 for 7(a) small loans at or below 350,000 dollars, a segment that had carried no numeric floor before (3)(4). SBA 504 is governed differently still, by a repayment-ability standard rather than a fixed numeric coverage floor.

Set those floors against the realized record and a tension appears, because the required coverage and the realized loss do not move together. The 504 program carries the lowest floor, 1.0, and posts the lowest seasoned charge-off rate, 1.49%, because its fixed-rate, fully collateralized, real-estate-backed structure does the work a ratio cannot. The small-loan segment that has only just received a numeric floor posts the highest loss rate, 7.27%, well above the standard 7(a) segment at 3.99% despite carrying a lighter requirement. The lesson the data carries into the new standards is the one that runs through this entire Index: a coverage ratio is a useful floor, but collateral, structure, and loan size are what actually separate the safest credit from the riskiest. The tightening is a measured response to a genuine deterioration, and because the floors took effect only in 2025 and 2026, their result will become visible in this record only as the post-2025 vintages season, which is precisely what a quarterly Index is built to observe.

4. Reading the data correctly: the seasoning problem

A naive reading of the file produces a comforting and false conclusion. Compute the charge-off rate by approval year and lending appears to have become dramatically safer after 2020, with recent cohorts charging off at a fraction of a percent. The reason is not safety. It is time.

A small-business loan that defaults rarely does so immediately. In this data the median charge-off arrives 4.7 years after approval, and roughly half of all charge-offs occur in years two through five, a timing pattern consistent with the discrete-time default-hazard tradition established by Glennon and Nigro, whose work found small-business default risk peaking in the early years of a loan's life rather than at origination (8)(9)(10). A loan approved in 2025 has had a few months, not a few years, and so it has not yet had the opportunity to fail. The share of each vintage still open makes the point directly: 1.5% of the 2010 cohort remains open today, against 96% of the 2025 cohort. As that open share rises toward the present, the measured charge-off rate falls toward zero, purely as an artifact of incomplete maturation.

The Index resolves this the way credible performance measurement must. All headline performance rates in this issue are computed on the fully seasoned fiscal 2010 through 2019 cohort, 478,961 funded 7(a) loans and 58,887 funded 504 loans that have had at least six years to resolve. Vintages from 2020 forward are reported separately and treated as still maturing. This is the line between a performance benchmark and a misleading snapshot, and it is why the numbers in this Index run higher, and truer, than a surface read of the same file would suggest.

5. The structural spread: 504 against 7(a)

The first substantive finding is also the thesis of the issue. Holding the industry constant, the program structure a sponsor selects moves the loss rate more than the industry itself does.

Across the seasoned cohort, the 7(a) program charges off at 6.44% and the 504 program at 1.49%. Sector by sector the ratio runs from roughly three to ten times. In transportation and warehousing a 7(a) loan charges off 10.5 times as often as a 504 loan; in professional services, 9.2 times; in construction, 8.7 times. There is no sector in which 504 performs worse.

The explanation is structural rather than mysterious. The 504 program finances fixed assets, most often owner-occupied real estate, through a first-lien third-party loan and a fixed-rate, fully amortizing SBA debenture in a second position, typically at lower combined leverage than a comparable 7(a) facility. Its credit standards reflect that conservatism through a repayment-ability test and collateral requirements rather than the kind of numeric coverage floor that governs 7(a), where standard loans must clear 1.15 to 1 and small loans 1.10 to 1 (3). Decades of mortgage-default research establish why the structure matters: origination debt-service coverage and leverage are among the strongest predictors of commercial-mortgage default, with default declining steadily as origination coverage rises (11)(12), a relationship documented specifically for lodging and other operating-asset collateral as well (13). The 504 structure is, in effect, a standing application of those findings, and the realized data shows the result.

For a sponsor weighing how to finance a real-estate-anchored project, this is not an academic point. It is the single largest lever available before the first dollar is drawn.

6. The base rate behind every feasibility conclusion

A feasibility study answers a question about a specific business in a specific place. The base rate that anchors that answer therefore has to be measured at the resolution of the business, not the sector, and the data shows why with unusual force.

Among the most active SBA industries, the seasoned adverse-resolution rate spans more than twentyfold, from 0.92% for self-storage to 20.63% for electronics stores. The decisive observation is that the spread inside a single two-digit sector frequently exceeds the spread between sectors. Within real estate and rental, self-storage charges off at 0.51% while residential property management charges off at 8.53%, a seventeenfold gap under one sector heading. Within accommodation and food, hotels charge off at 2.10% while limited-service restaurants charge off above 8%. Within health care, dental and optometry practices sit near 2.3% while home health care runs near 7%. A conclusion that an industry is "safe" or "risky" at the sector level is, at the resolution a lender actually prices, close to meaningless.

This is the analytical spine of the Index, and it is built to be used. The reference table reports each industry at the six-digit NAICS level, reconciling the 2017 and 2022 NAICS revisions so that a business type appears once rather than split across codes (16), and suppressing any industry with fewer than 500 loans so that no published base rate rests on a thin cell. The risk bands are anchored to the program-wide rate, so a flag marks an industry that is meaningfully worse than the SBA average rather than simply above zero.

7. The price of risk, and the Express effect

If structure is the first lever and asset class is the base rate, price is the most precise single signal in the file.

The relationship is close to monotonic. A 7(a) loan priced at or below 4% charges off at 2.94%; a loan priced above 11% charges off at 14.93%, roughly five times higher, and the rate rises in nearly unbroken steps across every band in between. Two qualifications make the signal sharper rather than weaker. The average loan size collapses from about 589,000 dollars in the lowest band to roughly 23,000 dollars in the highest, so the most expensive loans are also the smallest and most lightly collateralized. And the interaction with size is itself enormous.

At the intersection of size and price the charge-off rate runs from about 1% on large, low-priced loans to roughly 24% on small, high-priced ones, a twenty-four-fold range visible in a single grid, and within any size band the loss rate always rises with price. This is risk pricing functioning exactly as theory predicts, observed across nearly half a million loans.

The high-rate, small-loan tail is overwhelmingly one product, and it deserves its own treatment because the conventional intuition about it is wrong.

SBA Express is half of the 7(a) program by count, averages just 75,000 dollars against 685,000 dollars for standard 7(a), carries an interest rate a full point higher, and accounts for 97% of all loans priced above 9% (17). It is therefore the entire explanation for the high-rate tail. But it does not uniformly raise default. Split a given industry into Express and standard loans and the charge-off rates come out close to each other in most cases. The exception is precise and important: in capital-intensive businesses that normally rely on real-estate collateral, Express is far worse. A car wash charges off at 3.24% as a standard 7(a) loan and 9.89% as an Express loan; gas stations and convenience stores show the same penalty, because a small, lightly secured Express loan to a fixed-asset business is a structurally weaker proposition than the larger collateralized loan that business would otherwise carry. In labor-light trades such as landscaping, remodeling, and contracting, the small fast loan fits the business and Express performs better. The lesson for feasibility is direct: for a real-estate-anchored project, loan structure changes the odds, and the Express shortcut is a poor fit.

8. Capital against labor: jobs per dollar

The SBA's statutory purpose is job creation, and the data permits a question no incumbent benchmark asks: do the businesses that create the most jobs per dollar of public capital also survive?

They do not. The relationship runs the wrong way. Restaurants generate roughly 56 jobs per million dollars of 7(a) capital, the most labor-intensive use of the program, and carry one of the highest adverse-resolution rates at 11.08%. Self-storage generates 2.5 jobs per million dollars, the least labor-intensive use, and is the safest asset class in the entire book at 0.92%. Hotels, gas stations, and funeral homes cluster in the capital-intensive, low-risk quadrant; fitness centers and restaurants in the labor-intensive, fragile one. The businesses that do the most for the program's job-creation mandate are also the most likely to fail, and the capital-intensive, real-estate-anchored projects that create fewer jobs are the ones that endure. That tension sits at the center of federal small-business credit policy, and it is visible here in a million loans. It is also a finding a sponsor and a lender should weigh together, because the project that scores best on the program's stated purpose is not the project that scores best on survival.

9. The geography of default

Risk is not evenly distributed across the country, and the variation is wide enough to matter to a project-specific conclusion.

The seasoned adverse-resolution rate ranges more than threefold by state, from roughly 3.4% in the northern Rockies to about 12.4% in Florida. Florida is the highest-loss sizable state, followed by New Jersey, Louisiana, South Carolina, and Texas; the lowest-loss states are Montana, Idaho, and Vermont, with the Pacific Northwest and northern interior running consistently low. The pattern is not a proxy for any single factor, and the Index does not present it as one. It is a measured base rate by location, and at the county level within a target market the gradient is steeper still, which is where it becomes an input to a specific feasibility conclusion rather than a generalization about a state.

A note of discipline applies to all of these geographic figures. They describe where financed projects were located, conditioned on the loans the program actually made, and they are reported only for states with enough seasoned volume to support a stable rate.

10. The leading edge: where stress is building now

Everything above is historical base rate. The final exhibit is the reason the Index has a quarterly cadence, because it points at the next wave rather than the last one.

Active distress, the share of still-open loans that are delinquent, past due, or in deferment, is a leading indicator precisely because a charge-off arrives years after the trouble begins. On the open book the signal is concentrated in the 2022 and 2023 origination cohorts, running at 4.63% and 5.52%, the loans made into the sharpest part of the rate-shock window. The most recent cohorts, 2024 and 2025, read low only because they are too new to have surfaced distress, the same maturation effect the seasoning curve established. One caveat governs the interpretation of recent vintages: the agency's March 2026 Inspector General review (5) drives some recent resolution activity for administrative rather than purely credit reasons, so the distress signal, which captures borrower payment status directly, is the cleaner forward read. The 2022 and 2023 cohorts are the ones to watch, and the Index will track them as they age.

Methodology

Scope and source. This issue analyzes the complete public SBA 7(a) and 504 loan disclosures as refreshed to March 31, 2026, comprising 919,732 7(a) records and 116,345 504 records for approval years 2010 through the first quarter of 2026 (1)(2). This is a population, not a sample. Loan status is parsed from the data itself rather than the published data dictionary, which understates the disclosed status set (19).

Performance window. All charge-off, adverse-resolution, and base-rate figures are computed on the fully seasoned fiscal 2010 through 2019 cohort. Vintages from 2020 forward are reported only where their incompleteness is the point, as in the seasoning and early-warning exhibits.

Definitions. Charge-off is a loan in charged-off status as a share of funded loans. Adverse resolution combines charge-offs, liquidations, and guaranty purchases that are not charge-offs. Active distress combines delinquency, past-due, and deferment status as a share of loans still open. Funded loans exclude cancelled and undisbursed commitments.

Suppression and reconciliation. Industry base rates are suppressed below 500 funded loans; cross-tabulated cells below 30 loans are suppressed; state rates require at least 300 seasoned loans. The 2017 and 2022 NAICS revisions are reconciled so that a business type is reported under a single current code (16).

Conditioning. Where loan structure materially changes the reading, as with SBA Express, the analysis reports the conditioned result rather than a pooled average (17). The full charge-off, liquidation, and guaranty-purchase definitions follow the agency's loan-resolution conventions (20).

Limitations. The disclosure records loan status and resolution dates; it does not record borrower revenue, occupancy, or debt-service coverage, so this issue measures realized outcomes rather than the financial ratios that produced them. Recent-vintage resolution activity reflects administrative review alongside credit performance (5). A future issue will incorporate MMCG's own feasibility-engagement record alongside the public population, at which point convenience-sample disclosures will apply; they do not apply to this public-population issue.

Governance. The MMCG Feasibility Index voluntarily models its disclosure practices on the IOSCO Principles for Financial Benchmarks (2013) and is not itself a regulated financial benchmark (21).

About the author

Michal Mohelsky, J.D. is the Principal of MMCG Invest, LLC and lead author of the MMCG Feasibility Index. MMCG Invest is a national commercial real estate feasibility consulting firm that produces lender-grade, independent feasibility studies for SBA 7(a), SBA 504, USDA, and conventional loan programs across more than thirty asset classes.

About MMCG Invest, LLC

MMCG Invest is a feasibility study company that delivers third-party feasibility studies prepared to a lender's evidentiary standard, combining primary market analysis, asset-class benchmarking, and a proprietary data practice that includes the analysis presented in this Index. The firm's work is independent, methodologically transparent, and prepared for decision-makers who carry the risk.

Considering a project that will need a feasibility study a lender will accept? Book a Meeting.

Sources

U.S. Small Business Administration, 7(a) Loan Program FOIA dataset, data as of March 31, 2026.
U.S. Small Business Administration, 504/CDC Loan Program FOIA dataset, data as of March 31, 2026.
U.S. Small Business Administration, Standard Operating Procedure 50 10 8, effective June 1, 2025.
U.S. Small Business Administration, Procedural Notice 5000-876777, 7(a) Small Loan debt-service-coverage requirement, effective March 1, 2026.
U.S. Small Business Administration, Office of Inspector General, Report 26-07, March 11, 2026.
U.S. Small Business Administration, Office of Inspector General, Report 26-03, FY2025 financial statements, January 21, 2026.
U.S. Small Business Administration, FY2024 Agency Financial Report.
Glennon, D. and Nigro, P. (2005). Measuring the Default Risk of Small Business Loans: A Survival Analysis Approach. Journal of Money, Credit and Banking, 37(5), 923-947.
Glennon, D. and Nigro, P. (2005). Journal of Financial Services Research, 28(1), 77-111.
Glennon, D. and Nigro, P. (2011). Journal of Credit Risk.
Archer, W., Elmer, P., Harrison, D., and Ling, D. (2002). Determinants of Multifamily Mortgage Default. Real Estate Economics, 30(3), 445-473.
Ciochetti, B., Deng, Y., Lee, G., Shilling, J., and Yao, R. (2003). Journal of Real Estate Finance and Economics.
Singh, A. (2019). Journal of Hospitality and Tourism Research, lodging CMBS default and origination debt-service coverage.
Federal Reserve Banks (2025). Small Business Credit Survey, employer-firm results.
Yale School of Management (2025). SBA 7(a) program teaching case, February 5, 2025.
U.S. Census Bureau, North American Industry Classification System, 2017 and 2022 revisions.
U.S. Small Business Administration, SBA Express program terms, guaranty percentage and interest-rate maximums.
U.S. Small Business Administration, FY2024 and FY2025 lending activity reports.
U.S. Small Business Administration, 7(a) and 504 loan data dictionary, loan-status field.
U.S. Small Business Administration, loan liquidation, guaranty-purchase, and charge-off processing guidance.
International Organization of Securities Commissions (2013). Principles for Financial Benchmarks, IOSCOPD415.

Lets Discuss Your Project

Every project has its own program, lender, and documentation context. Book a working session and we will review the site, the comparable set, and the lender requirements together. No prepared pitch, no obligation.

Engagements are led by Michal Mohelsky, J.D., Practicing Affiliate of the Appraisal Institute. Feasibility studies are prepared under USPAP discipline, aligned with SBA SOP 50 10 8 for 7(a) and 504 loans and with 7 CFR Part 5001, Appendix A to Subpart D for USDA Business and Industry, REAP, and Community Facilities financing. Engagements start at $4,900 with fixed-fee scoping. Standard delivery is 9 to 16 business days, with rush turnaround available from 5 days. A senior analyst responds to proposal requests within 12 business hours from the firm's San Francisco office at 27 Maiden Lane, Suite 625.