The Approval-Cohort Charge-Off Floor: Why a Published SBA Loss Rate Can Only Understate

A loss rate for SBA loans looks like the plainest number in small business credit: loans that failed, divided by loans that were made. In practice it is a stack of choices. The answer depends on which loans stand in the denominator, how long they have been on the books, which outcomes the publisher is permitted to disclose, and whether loans or dollars are being counted. Change one of those choices and the same loan file yields a different figure, sometimes several times larger or smaller. This article sets out the convention that the Loan Analytics loan dataset applies to every charge-off rate it prints, the approval-cohort floor, and explains why a rate built that way is a lower bound: each new release of the data can push it up, and no release can push it down short of a correction to the record.

The headline figures make the point. Take every disbursed 7(a) loan whose approval fell between fiscal 2017 and fiscal 2021: by September 30, 2026, 6.33 percent ^[1] of them had been charged off. For 504 loans from the same approval years the figure is 0.35 percent ^[2]. Both are floors. What follows explains what that word commits the reader to, how the dataset fixes the horizon so that cohorts of different ages can be set side by side, why the dollar rate reads differently from the count rate, why the two programs cannot be compared at face value, and how a feasibility study should and should not use a floor. Two archived posts approach the same file from the outcome side. SBA Loan Default Rate by Industry: Charge-Off Rates for 416 NAICS Codes ranks industries on an older and largely finished cohort, and The Next SBA Default Wave Is Already in the Data. It Just Doesn't Look Like One. reads recent vintages at matched ages. This article is about the measuring rule that sits underneath both.

What an approval cohort is, and why the rate is not taken by calendar year

An approval cohort is the group of loans SBA approved in one fiscal year, or in a run of fiscal years, followed forward through time as a unit. Membership is settled on the day of approval and never changes afterwards. A loan approved in fiscal 2018 belongs to the fiscal 2018 cohort whether it pays off in three years, runs to maturity, or is written off in its seventh year. Each later release of the loan file can change what is known about the members, but it cannot add a member or drop one. That fixed membership is what lets a cohort rate be read at all: its numerator and its denominator describe the very same loans.

The alternative is a calendar-period rate, and most readers meet that convention first, because bank regulators publish losses that way. The Federal Reserve's series on commercial bank charge-offs takes a bank's net charge-offs during a quarter, gross charge-offs minus recoveries, and divides them by the average level of loans outstanding over the quarter ^[3]. The quarterly ratio is then scaled up so that it reads as an annual rate ^[4]. A calendar rate answers a question about a book: how much did this portfolio lose this year, against its size this year? For a bank managing earnings and reserves, that is a sensible question.

For anyone trying to learn how a kind of loan performs over its life, a calendar rate blends two things that ought to be kept apart. The losses recorded in a given year come mostly from loans made several years before, while the loans outstanding in that year include everything made recently, most of which has not yet had time to go wrong. When lending volume grows, the denominator fills with young loans faster than the numerator can catch up, and the calendar rate drifts down even if credit quality has not moved at all. When volume shrinks, the drift runs the other way.

SBA's own volume has moved enough for that drift to matter. The program approved 42,298 ^[5] 7(a) loans in fiscal 2020 and 78,078 ^[6] in fiscal 2025, so any rate whose denominator is the book outstanding in a recent year leans heavily on loans only a year or two old. A cohort rate avoids that distortion by construction. It asks what has happened, up to a stated date, to a fixed group of loans whose age is known.

Bank supervisors use the same idea under another name. The 2020 interagency policy statement on credit loss allowances lists vintage analysis among the loss-rate methods a bank may apply under the current expected credit loss standard ^[7], and it names vintage, alongside industry of the borrower, collateral type and term, among the traits by which loans may be pooled for estimation ^[8]. An approval cohort is simply a vintage keyed to the SBA approval date.

The approval date, rather than the date money first went out, anchors the cohort because it is the date on which SBA counts its own lending, so a cohort defined by approval fiscal year ties back to the program's yearly totals. Congress used the same anchor when it wrote the pandemic payment relief for SBA borrowers and made eligibility turn on the date SBA approved each covered loan ^[9]. Approval carries one complication. Some approved loans never receive any money: they are cancelled, or at the as-of date they are still waiting for a first disbursement. A loan that was never funded cannot be charged off, so the dataset leaves those loans out of the charge-off denominator and measures them separately, through a cancellation rate that the companion article on cancellation explains. In this dataset a charge-off rate always means charge-offs over disbursed loans.

Seasoning: how old the FY2017 to FY2021 cohort is

A cohort rate means little until the reader knows the age at which it was taken. Seasoning is the time a loan has spent on the books, and for a cohort it is a span rather than one number. The charge-off cohort in the Loan Analytics dataset runs from fiscal 2017 through fiscal 2021. Fiscal 2017 opened on October 1, 2016, and fiscal 2021 closed on September 30, 2021. Read on September 30, 2026, the as-of date of the SBA files, the oldest approvals in the pool are ten years old and the youngest five, and every charge-off figure the dataset publishes for this pool states that span in its definition.

Five to ten years sounds like plenty until it is set against the terms these loans carry. Among other limits in 13 CFR 120.212, a 7(a) term is capped at ten years except where the money pays for real estate, or for equipment expected to last more than ten years ^[10]; even then the ceiling is 25 years with extensions counted, and only the real property part may add the time needed to finish construction ^[11]. SBA offers 504 loans at 10, 20 and 25 year maturities ^[12]. A real estate loan approved in fiscal 2021 has, by the as-of date, worked through only a small part of its schedule. Its borrower may be prospering, struggling without showing it, or deep in a workout that has not yet produced a final entry. None of those conditions appears as a charge-off until the lender and SBA have finished with the loan. The write-off is the closing line of a loan's record, so the date it carries can trail the onset of trouble by a long interval.

Table 1 shows what age does to the 7(a) floor. Each row is one approval year, read on the same date, so each row is a year older than the row beneath it. The first column is the charge-off rate to date. The fiscal 2010 cohort, now 16 to 17 years old, has charged off 9.19 percent ^[13] of its disbursed loans; the fiscal 2021 cohort, at 5 to 6 years, has charged off 3.20 percent ^[14]. Part of that gap may reflect a real difference between two groups of borrowers who faced different economies; the file records outcomes, not their causes. Part of it is only that the fiscal 2021 cohort has not lived long enough to fail in the numbers its elders did. The remaining columns pull the two apart. For each cohort old enough to have been watched for seven years or more, between roughly two fifths and one half of the charge-offs recorded so far are dated after the fifth anniversary of approval. For the fiscal 2021 cohort the share is far smaller, because its loans crossed that anniversary only during the final year of the window. Most of the late losses that the older rows already contain still lie ahead of it.

The 504 program shows the same effect more sharply (Table 2). The fiscal 2010 504 cohort has charged off 3.34 percent ^[15] of its disbursed loans, and the fiscal 2021 cohort 0.15 percent ^[16]. The to-date rate falls nearly year by year as the cohorts get younger, so anyone who took the youngest 504 cohort's figure as the program's loss record would miss most of what the oldest cohort has already written off.

This is the first reason a published cohort rate understates. Until a cohort has run off entirely, some of the losses it will eventually record have not occurred yet, or have occurred in substance without yet reaching the file. Pooling five approval years into one cohort, as the dataset does for fiscal 2017 through 2021, smooths the noise between single years and lifts the denominator far above the size at which small numbers dominate, but it leaves the logic unchanged. The pooled figure is an average of rates that are each still climbing.

Charge-offs to date over disbursed loans: where the floor comes from

The arithmetic behind the floor is short. The numerator counts the loans in the cohort whose status in the SBA file reads charged off on the as-of date. The denominator counts the loans in the cohort that were disbursed, which is every loan that received money, whatever has happened to it since. In the current file a disbursed loan sits in one of a few states: paid in full, charged off, closed in some other way (a small group found only in the 504 file), or open.

The open loans are the source of the floor. SBA's disclosure regulation treats a loan's status as generally exempt from release unless the loan has been charged off or paid in full, and it treats non-statistical detail on defaults, delinquencies and losses the same way ^[17]. In the September 2026 release every open loan carries one undifferentiated withheld status, so the file cannot say whether an open loan is current, behind, deferred or in liquidation. In the 7(a) cohort, 68,103 ^[18] open loans sit inside the denominator of the charge-off rate; in the 504 cohort the count is 22,353 ^[19].

Those open loans enter the denominator as loans that have not been charged off. That is accurate on the date of the file and silent about the future. Some will pay in full. Some will be charged off in a later release. When one is written off, the numerator gains a loan and the denominator stays where it was, because the loan was in it all along. As long as the cohort's membership holds and its statuses are not revised, the rate can move in one direction only, and that direction is up. SBA refreshes the FOIA files every quarter, with each release normally posted about a month after the quarter closes ^[20], and each refresh can add charge-offs to an old cohort; only a correction to the record could take one away.

That is why the dataset labels every charge-off rate a floor and prints the count of open loans beside it. The label is not a hedge or a disclaimer. It is a statement about the direction of every future revision, and a reader can test it by comparing the same cohort across two releases.

Loan Data Behind This Article

Every figure in this table comes from the Loan Analytics loan dataset, analytics-loan-data freeze-3 (commit 360c856), and carries its key. SBA rates are approval-cohort floors and are withheld when fewer than 30 loans stand behind them.

Why the open loans stay in the denominator

Why keep open loans in the denominator at all? The tempting alternative is to drop them and compute the rate on resolved loans only, those that have reached a final status. That figure is larger, and it can feel more candid, since it seems to answer how many of the finished loans failed. But loans do not finish at random. In a cohort aged five to ten years, the loans already resolved lean toward those that resolved early: businesses that sold or refinanced, shorter working capital loans that ran their full term, and loans that failed fast. The loans still open lean toward long real estate loans paying on schedule, along with an unknown number in trouble that the file cannot show. A resolved-only rate weights the cohort toward its quickest outcomes, good and bad alike, and nothing in the file reveals which way that bias runs for a given asset class.

Table 4 prints both versions for two asset classes in each program, the full-service restaurant and the hotel and motel cells, and adds two lines that turn the floor into something concrete. One is the floor's sensitivity to the open loans: the number of points the rate would gain for each one percent of open loans that later charge off. The other is the arithmetic ceiling, the rate that would follow if every open loan were written off. Nobody expects that outcome, and the line is there only to show that the published floor has no useful bound above it. The cohort's eventual rate will land somewhere between the floor and that ceiling, and the file by itself cannot narrow the range.

Take the 7(a) restaurant column. The cohort's 12,099 ^[25] disbursed loans include 3,552 ^[25] that are still open and 807 ^[25] that have been written off, which yields the class floor of 6.67 percent ^[25]. Because the open loans make up a large slice of the denominator, even a modest share of them going bad in later years would lift the published rate by a visible margin, and the sensitivity line gives the exact increment for each point of that share.

Holding the horizon fixed: the within-5-years rate

A floor that keeps rising is awkward to compare. Two cohorts read on the same date differ in age, so the older one has had longer to collect charge-offs, and the gap between their to-date rates mixes credit quality with elapsed time. The dataset's answer is a second rate that pins the horizon: the share of a cohort's disbursed loans whose write-off is dated no later than the fifth anniversary of approval. Every loan in the fiscal 2017 to 2021 pool is at least five years old on September 30, 2026, so for every member the five-year window has closed, and the measure covers the same stretch of each loan's life.

For the pooled cohort, the 7(a) within-5-years rate is 3.72 percent ^[21], against the to-date floor of 6.33 percent ^[1]. For 504 it is 0.17 percent ^[23], against 0.35 percent ^[2]. In each pair, the distance between the two numbers is the part of the recorded loss that arrived after year five, and that part exists only for loans old enough to have lived past their fifth anniversary. The fixed-horizon rate surrenders it on purpose, in exchange for a figure that means the same thing in every cohort.

That trade is what the second column of Table 1 buys. Read down the to-date column and the 7(a) cohorts appear to improve and then deteriorate in a pattern that partly tracks their age. Read down the within-5-years column and a cleaner picture emerges. The fiscal 2017 and 2018 cohorts wrote off more of their loans inside five years (4.30 percent ^[29] and 4.65 percent ^[30]) than any cohort from fiscal 2011 through 2016, whose rates run from 3.07 percent ^[31] to 3.86 percent ^[32]. That is a statement about those cohorts at an equal age, which the to-date column, taken alone, cannot make. The 504 table tells a similar story in its own column: the 504 cohorts of the early 2010s wrote off more of their loans inside five years, 1.43 percent ^[33] for fiscal 2010, than any 504 cohort approved after fiscal 2012, and a fixed horizon assigns that difference to the cohorts themselves rather than to their age.

The fiscal 2020 and 2021 cohorts show the lowest within-5-years rates in the 7(a) table, 2.58 percent ^[34] and 2.60 percent ^[35]. A fixed horizon removes the effect of age, but not the effect of what happened during those five years. In the pandemic, Congress directed SBA to pay the principal, interest and fees owed on covered loans in regular servicing for set periods ^[36], and the covered loans included those made under the 504 statute as well as 7(a) loans ^[37]. For a covered loan made between late March and late September 2020, the statute directed SBA to make the first six months of payments ^[38], and it did the same for a covered loan approved between February and September 2021 ^[39]. Older loans in regular servicing received relief too, but the fiscal 2020 and 2021 cohorts received it at the very start of their lives, which is when a new loan's early failures would otherwise show. Their low early charge-off rates may owe as much to that relief as to underwriting, and the file cannot separate the two. A study that compares cohorts on the within-5-years rate should say so, and should not read the pandemic cohorts' light early losses as proof of better credit.

The fixed horizon has one more limit. The window counts charge-off dates, not default dates. A loan that stopped paying in its fourth year but was written off in its sixth counts as an after-year-five loss. If the time from default to write-off changed across the years in the table, because of liquidation backlogs, servicing policy or the mix of collateral, the within-5-years rate would shift for reasons unrelated to credit. The limit may weigh more on 504 loans, whose piece of the financing normally ranks second in its claim on the property ^[40], so its outcome depends on how and when the property is dealt with. For the youngest cohort, whose window closed only in the last year of the file, a write-off recorded late but dated inside the window can still nudge the figure. The within-5-years rate is the fairest comparison the public file allows, not a perfect one.

The dollar rate: written-off balances against approved dollars

A count rate gives a small working capital loan and a large real estate loan one vote each. To show the other view, the dataset prints a dollar rate for each cohort: the gross charge-off amount on written-off loans, divided by the gross approved amount of every disbursed loan. For the 7(a) cohort it is 1.95 percent ^[22], well under the count floor; for 504 it is 0.29 percent ^[24].

Two forces can hold a dollar rate below a count rate, and the file does not separate them. The first is loan size. If smaller loans are written off more often than larger ones, the written-off loans carry less than their share of approved dollars. The second is repayment before failure. The charge-off amount is the balance actually written off, and it is set against the full original approval, so a loan that had repaid part of its principal before it failed adds less than its approval to the numerator. Either force, or both, can be at work in any cell. A study that quotes the dollar rate should describe it as a ratio of written-off balances to original approvals, not as the share of money that was at risk when trouble began.

The word gross matters as well. The charge-off amount is gross of recoveries, so money collected after the write-off does not reduce it. It is also gross of the guaranty. A 7(a) loan of $150,000 or less can carry a guaranty of up to 85 percent, and a larger loan up to 75 percent ^[41], so the written-off balance is shared between the lender and SBA, and neither party's loss equals the figure in the file. The Federal Reserve's bank charge-off rates, by contrast, are net of recoveries ^[3]. A gross cohort floor and a net calendar rate are different instruments, and setting one against the other says more about definitions than about borrowers. The SBA file carries no recovery field and no measure of loss severity, so the dataset prints neither, and the series article on the limits of the file treats that gap at length.

7(a) against 504: two floors that do not sit on one scale

The two programs' floors differ by more than an order of magnitude, 6.33 percent ^[1] for 7(a) against 0.35 percent ^[2] for 504. It is natural to read that gap as a ranking of credit risk, and some of it may be. The structures differ in ways that could plausibly bear on credit. A 504 project is financed in layers: one or more third party loans at least as large as the 504 loan ^[42], a 504 loan that normally stands second in line, behind the first lender, in its claim on the project property ^[40], and the borrower's own cash or land, worth no less than 10 percent of project cost in the ordinary case and 20 percent where a young business takes on a limited or single purpose property ^[43]. SBA describes the program as long-term, fixed rate financing for major fixed assets ^[44]. Under 7(a), inventory and working capital can be financed too ^[45], alone or alongside real estate and equipment, with a term of ten years or less unless real estate or long-lived equipment is financed ^[10]. The companion article on the 504 method sets out the structure in full.

Structure is not the only difference, though, and the others come from measurement rather than credit. The first is the open share. 504 loans carry maturities as long as 25 years ^[12], and at five to ten years most of them are still on the books. In the restaurant and hotel cells of Table 4, the open share of the 504 denominator is 77.6 percent and 71.7 percent, against 29.4 percent and 38.0 percent for 7(a). With so much of the denominator unresolved, the 504 floor is far more provisional than the 7(a) floor: every point of open 504 loans that later goes bad adds more to the rate, as the sensitivity line of the table shows.

The second is the timing of loss. Table 2 shows the 504 to-date rate climbing steadily with cohort age, much more steeply in proportion than the 7(a) rate does in Table 1. If 504 losses tend to arrive later in a loan's life, a window of five to ten years captures a smaller fraction of them, and the pooled 504 floor understates its cohort's eventual losses by a wider margin than the 7(a) floor does. The within-5-years rates do not repair this. They compare the programs over the same early span, and that early span is exactly when a 504 loan is least likely to have reached any final status.

The third is the mix of businesses. 504 money goes to fixed assets, and the industries that lean on the program differ from those that lean on 7(a). Setting the two programs side by side inside one asset class removes that mix effect, which is why the restaurant and hotel cells stand next to each other in Table 4. Among full-service restaurants, the 7(a) floor is 6.67 percent ^[25] and the 504 floor 0.94 percent ^[26]. Among hotels and motels, the 7(a) floor is 0.99 percent ^[27] and the 504 floor 0.55 percent ^[28]. The gap survives inside each class, which shows the program difference is not purely an industry effect. It does not show how much of the gap is structure, how much is the selection of borrowers into each program, and how much is the longer life of a 504 loan, and nothing in the file can divide it up.

A study that cites the 504 floor for a 504 project should therefore call it a floor on a largely unresolved denominator and should not offer it as evidence that the project is many times safer than a 7(a) project of the same kind. A program's design shapes who borrows, what they buy, how long they owe and who stands first in line, and every one of those also shapes when, and whether, a loss shows up in the record.

How a feasibility study uses a floor

A feasibility study for an SBA loan answers a narrow question: whether the revenue a single project can earn, after its costs, will service the debt a lender is being asked to extend. The agency's rule on loan conditions lets it ask for a study, alongside appraisals or a survey ^[46]. Loss data enters a study as background to that question and never as the answer to it. A floor is useful background precisely because the direction of its error is known. It says that at least this share of comparable loans, approved in these years and read at this age, has already been written off.

Four uses follow from that, along with one misuse a study has to avoid.

As a lower bound on program or class experience. When a study describes the lending record for an asset class, it can give the class floor and name, right beside it, the approval years, the age span and how many loans remain open, and say plainly that the eventual write-off count for that cohort will be at least that high. That is a true statement, and a reviewer can check it against the key printed beside the figure. The 7(a) restaurant floor of 6.67 percent ^[25] is one such statement: no later release of the file can show fewer restaurant loans from those years written off, short of a correction to the record or a redrawn class.

As a relative measure. The ratio of a class floor to its program's floor, or of one class to another inside the same program and cohort, often tells a lender more than either level does, because both sides share the same approval years and the same disclosure rules. The match is not exact: a class whose approvals cluster in the later years of the window is younger on average than its program, and its to-date floor should then be read beside its within-5-years rate. A class whose floor sits well above its program's says that loans of that kind have done worse than the program overall, at least so far. When the comparison runs across cohorts rather than across classes, a study uses the within-5-years rate and not the to-date floor.

As a dated observation. Because every quarterly release can raise a floor ^[20], a study names the release behind each rate it prints (for this article, the September 30, 2026 files), and a reader who later opens a newer release then knows why the number moved and which way it ought to have moved.

As a prompt for the stress case. A study's sensitivity analysis asks what happens to debt service coverage when revenue, margins or occupancy fall short of the base case. A high class floor does not set the size of those shocks, but it is a reason to test them hard, and to explain why the project's own evidence supports the base case regardless.

Conventions. This article follows the conventions of the Loan Analytics worked examples. Coverage is tested against the program floor the site's program page states, where the program has one, and against a stated 1.25x lender assumption, and the higher of the two binds. Equity is measured at the loan the cash flow supports, with fees recomputed on that loan. An acquisition is tested on the seller's historical earnings, not on projections.

Never as a probability of default for one project. A floor is a tally of outcomes in a pool of loans approved years ago, to borrowers the file does not describe, on terms it shows only in part. It is not the chance that a particular borrower, with a particular plan, site, equity stake and lease, will fail on a loan approved today. Even a fully resolved cohort rate would not be that. It would be the average outcome of a mixed group under the economic conditions of its own years. Bank accounting draws the same line. Under the current expected credit loss standard, as the 2020 interagency policy statement describes it, a bank measures expected losses over the contractual term of its loans, allowing for prepayments ^[47], and adjusts historical loss information for differences in underwriting standards, portfolio mix and terms between the historical loans and the loans being assessed ^[48]. A cohort floor read at five to ten years is neither a lifetime figure nor adjusted to any particular loan, and a lender sizing a reserve and a study describing context both have to say what the historical number leaves out.

Two cautions sit beside these uses. The first is that a floor of zero is still only a floor. The 7(a) RV park and campground cell shows a floor of 0.00 percent ^[49] on 274 ^[49] disbursed loans, 110 ^[49] of them still open. That means no loan in the cohort has been written off yet; it does not mean the class carries no risk, and a study that read it that way would be repeating the young-cohort error at the level of an industry. The second concerns thin cells. A class or state cell with few disbursed loans can show a high floor because one loan failed or a low one because none has yet. Rates that rest on fewer than 30 loans are left blank in the dataset, and the companion article on small samples explains that rule and what a study does instead.

What to check in any published SBA loss rate

The same questions apply to any SBA loss figure a lender or borrower is handed, whether it comes from this dataset, an archived post, a vendor or a lender's own presentation. Each question has a short answer when the rate is well documented, and a missing answer is itself informative.

• Cohort or calendar. Does the rate follow a fixed group of loans, or divide one period's losses by one period's book, as the Federal Reserve's bank series does ^[3]? The two cannot be compared without restating one of them.
• Which denominator. Approved loans, disbursed loans, loans outstanding, or resolved loans only? Each gives a different number from the same file, and a resolved-only rate has a bias whose direction is unknown.
• What age. How many years separate the approval dates from the as-of date, and does the source say? A rate with no stated seasoning cannot be placed on the curve in Table 1.
• Which release. SBA updates the files quarterly ^[20], and the status vocabulary has changed between releases, a change described in the archived post The Next SBA Default Wave Is Already in the Data. It Just Doesn't Look Like One. A rate without a release date cannot be reproduced.
• How many open loans. If the source does not print the open count inside its denominator, the reader cannot judge how provisional the rate is. The dataset prints 68,103 ^[18] for the 7(a) cohort for that reason.
• Count or dollars, gross or net. A count rate and a dollar rate answer different questions, and a gross write-off is not a net loss.
• Which program. A blend of 7(a) and 504 loans mixes two denominators with very different open shares, and the blend moves with the program mix alone.
• To date or fixed horizon. Comparisons across approval years need a fixed horizon; a to-date floor favours the youngest cohort every time.
• How many loans. A rate built on a handful of loans is mostly noise, which is why this dataset prints no rate whose denominator falls below 30 loans.
• What cause is claimed. The file records outcomes, not reasons. A source that explains why a class failed is reporting something the loan file cannot show.

The archived post SBA Loan Default Rate by Industry: Charge-Off Rates for 416 NAICS Codes answers the seasoning problem the other legitimate way: it waits, using an older 7(a) cohort most of whose loans have reached a final status. Its rates sit closer to lifetime figures and say less about recent approval years. The floor in this dataset covers recent approval years and sits further from a lifetime figure. Neither convention is wrong, and the two cannot be compared until both are stated. Loan Analytics prints its convention beside every rate for that reason.

The inputs behind the tables

Every number in the two cohort tables and in Table 4 is computed from the inputs below, each read from the dataset under its key and cited to it. The cohort rates enter at two decimals, as the dataset displays them, so a difference or a share printed in a table can depart by a unit in the last place from the same calculation on unrounded values. The cell counts enter exactly.

Rounded as shown; lines may not foot by one unit in the last place.

Related Pages on Loan Analytics

• The Loan Analytics methodology series (all articles)
• How a feasibility study is done: the Loan Analytics methodology
• SBA feasibility studies (7(a) and 504)
• Loan data by state
• Loan data by asset class
• Worked examples (case studies)
• Worked Example: SBA 7(a) Assisted Living Feasibility Study in Utah
• Worked Example: SBA 504 Express Car Wash Feasibility Study in Maryland
• Worked Example: SBA 7(a) Gas Station Feasibility Study in Tennessee
• Worked Example: SBA 7(a) Gym Feasibility Study in Hawaii
• Worked Example: SBA 504 RV and Boat Storage Feasibility Study in New Jersey
• Worked Example: SBA 504 Owner-Occupied Retail Feasibility Study in Michigan
• SBA Loan Default Rate by Industry: Charge-Off Rates for 416 NAICS Codes
• The Next SBA Default Wave Is Already in the Data. It Just Doesn't Look Like One.
• Why a Rate on Fewer Than 30 Loans Is Withheld: Small Samples and the Suppression Rule
• Cancellation Is Not Charge-Off: The Two Ways an SBA Approval Can End Without Repayment

Commission a Feasibility Study

Loan Analytics prepares lender-grade feasibility studies for SBA 7(a), SBA 504, USDA Business and Industry, conventional and EB-5 financing, built on the methods this series sets out. A study is not an appraisal: it tests whether a project's market, costs and cash flow can carry the debt a lender is asked to approve. To discuss a study for your project, contact Loan Analytics.

Daniel Smith, MAI, Loan Analytics