Commercial financial spreading, explained.

Financial statement spreading explained: a business, its related companies and its owners on one template

One borrower package, from 38 documents to committee

1. The borrower sends a packageArnevik sends 38 documents: three years of returns for two companies and two households, 12 K-1s, reviewed statements, an August interim.
2. The relationship manager checks it is completeThe last K-1s arrive at 4:40 p.m. on Friday. The memo is due Wednesday.
3. An analyst spreads itEvery figure into the bank's template, tied out, combined into global cash flow. This is where the double-counted $591,000 is caught, or carried forward.
4. A senior underwriter signs, and a memo is writtenThe relationship is over the analyst's $1,000,000 limit, so a second signature is needed.
5. Committee decides, and the spread comes back every yearAnnual reviews repeat the whole job. A spread kept in a workbook starts again from a blank template.

Step 3 is the subject of this guide. The people, companies and figures in this example are fictional.

Why taxable income misleads, and three more spreading surprises

Spread, K-1, global cash flow, debt service coverage

Fourteen terms are defined in the one-page summary.

What financial spreading means: one template, every borrower, every line mapped

In shortA spread restates each borrower's figures in the bank's own lines so that every borrower is measured the same way. The work is mostly mapping: deciding which template line each figure belongs on, and keeping that decision the same next year. Bank spreads also depart from accounting rules on purpose, to show cash a lender can count on.

A bank's spreading template is a chart of accounts for borrowers: a fixed list of lines, each with a code, for sales, cost of sales, officers' pay, rent, depreciation, cash, receivables, inventory, debt and the rest. Ratios, covenants and risk ratings are built on those lines, so two analysts who spread the same borrower should get the same numbers. The Federal Reserve's examination manual expects the credit file to let a reviewer see the "income statement, balance sheet, reconciliation of equity, cash-flow statements, and applicable notes".3

Three things an analyst does to every line

Why bank spreads depart from accounting rules

Accounting statements show what a business owns and earned under generally accepted accounting principles. A lender wants to know what will turn into cash to pay the loan. The RMA guide, now published by ProSight, lists the standard departures: current assets that are restricted, that will not turn into cash within a year, or that are owed by related parties are spread as non-current, and long-term debts owed to related parties are spread as current.1,6

A loan to the owner sitting in "other receivables" is a common example, and many banks take such amounts out of tangible net worth as well.

What one owner-managed package holds

Three years of returns for two companies and two households, with a K-1 for each owner of each company each year, make 24 documents before a single accountant's statement. Owner-managed credits often look like this, which is why the number of entities sets the length of a spread more than the size of the loan does.

The rules that shape a commercial spread, and what changed in 2026

In shortNo federal rule says how to spread. Bank regulators require each bank's own lending policy to set what statements it needs and how it analyzes them, then examine the bank against that policy. SBA is the exception, with exact floors and documents. Five texts changed in 2026: the model guidance, SBA's SOP, the OCC's lending booklet, the small business data rule and Colorado's law on automated decisions.

SBA's numbers, as of October 2026

Audited, reviewed, compiled or tax return: what each document in a package can bear

In shortAn audit tests the numbers against outside evidence, a review asks questions and compares, a compilation puts the owner's numbers in proper form, and a tax return is what the business told the IRS, which a lender can check. The interim is usually the least checked and the most current. A careful spread ties each source to the others.

No federal rule sets one standard for every commercial loan. The examination manual tells examiners that "Much of the financial information that examiners will review will not be audited", and asks the bank to obtain at least annual statements.3

Program rules mostly leave the level to the lender. USDA's current business and industry guarantee rule says the lender "will determine the type and frequency of submission of financial statements", with at least annual statements prepared by an accountant, and lets the agency ask for annual audited statements when circumstances warrant.18 SBA, for acquisitions, now asks lenders to use "the highest level of financial reporting available".2

Most business returns are pass-through returns

A pass-through company pays no income tax itself. Its profit passes to its owners on K-1s and is taxed on their personal returns. That is why an owner-managed credit brings so many forms, and why the same profit appears on the company's return, the owner's K-1 and the owner's Schedule E.

The schedules that bridge the books and the return

Reading a K-1 for a loan: taxable income on one line, cash on another

In shortA K-1 carries two kinds of number. The income boxes are the owner's share of taxable profit, whether or not any cash moved. The distribution box is the cash the owner was paid. A lender who already counts the company's profit uses the income boxes as a check and counts the cash only under a method that leaves the company out.

IRS shareholder's and partner's instructions for Schedule K-1.20,21 Box 19 code A covers "distributions the partnership made to you of money"; distributions reduce the owner's basis and are generally taxed only beyond it.21,20

One property company, three readings

In our fictional example, Arnevik Properties owns the plant and leases it to the operating company for $504,000 a year. In 2025 its owners carried $286,000 of income from it onto their Schedule E. Its K-1s show $96,000 actually paid out. The other $190,000 was taxed to the owners but never paid to them.

The commonest error is mixing the readings: counting the property company's own income in global cash flow, then adding the owners' Schedule E income on top. Last year's spreadsheet in our example did exactly that with both companies, $591,000 in all. The second error runs through the operating company's distributions: $600,000 paid to the owners in 2025, $420,000 and $180,000 on their K-1s, came out of the same profit already counted as the company's cash flow.

Distributions against profit: an early warning

Distributions above net income mean equity is leaving the business. Arnevik Precision Machining paid out 138%, 215% and 197% of its net income over three years, and in the first eight months of 2026 it paid $360,000 while losing $69,000. Some of that is cash to cover the owners' tax on income that stayed in the company, which SBA names as a line a lender must justify.2 The rest is equity leaving a company that is asking to borrow.

How to calculate global debt service coverage without counting a dollar twice

In shortAdd up the cash flow of each company once, remove the money that only moves between them, add the guarantors' income from outside those companies, take off their taxes and living costs, and divide by every loan payment the group owes, the new loan included. The arithmetic is short. Keeping it honest across four parties and three years is the work.

The definition most banks quote comes from the 2023 interagency workout statement: cash flows of "both the borrower(s) and guarantor(s)" against "the combined financial obligations (including contingent obligations)".7 SBA asks for the same idea and names the lines that cause trouble: expenses and distributions, distributions for S corporation taxes, rent payments and owner's draw.2

A worked example, fiscal 2025

Fictional example. The new loan is $2.8 million over seven years at 7.25%: $42,602.52 a month, $511,230 a year. Three lines are missing on purpose: the owners' Schedule E income, the operating company's distributions, and the owners' share of the property company's income. All three are the companies' cash seen again from the owners' side.

What each common slip does to the same borrower

Every error on the chart moves the ratio the same way, up. That is why a second signer wants to see where every number came from, and why a spreadsheet template gives little protection on its own: it accepts whatever is typed into it, from whichever box.

There are two honest methods, and one error

Mixing the two in one spreadsheet is the error. So is counting cash from a company without its K-1, or keeping a personal mortgage out of the denominator while counting the rent that pays it.

What a good ratio is: turn it into dollars

No regulation names a good global ratio, only floors: SBA's 1:1 globally and 1.15 on the business, and each bank's own, such as the example bank's 1.25x for the group and 1.20x for the business.2 A more useful reading is the cushion: the cash the group could lose before it reaches the floor.

On $18.64 million of sales, the $346,000 cushion is about 1.9 points of gross margin. Then ask what could take the cushion away. In this case, the August interim shows margin already 3.4 points lower.

Interims, trailing twelve months and the ratios a credit committee reads first

In shortThe fiscal year can be ten months old by the time a committee meets, so the current year rests on the interim. Build the trailing twelve months from the fiscal year plus this year's interim less last year's same interim. Read business coverage beside global coverage, because they can move in opposite directions.

The examination manual names the mistake this avoids: "Mixing annual financial data with interim financial information can cause misinterpretation of cash flow."3 SBA's acquisition rules now ask for "the most current interim financial statement available along with the comparable interim financial statement from the previous year".2 In the example, fiscal 2025 sales of $18,640,000 plus $12,780,000 for eight months of 2026, less $12,240,000 for the same months of 2025, give trailing sales of $19,180,000 to August 31, 2026.

The two ratios answer different questions. Global coverage says the family can carry the debt; business coverage says whether the operating company can carry its own without the owners' wages and the property company's rent behind it. A committee that reads only the global figure would miss that the company now leans on its owners. Gross margin fell from 25.2% to 21.8% over the eight months, after supplier surcharges on a fixed-price contract, and inventory rose 32% to $2,840,000.

Why EBITDA misses cash tied up in inventory

Most global worksheets start from EBITDA, close to what the examination manual calls the traditional method: net income with non-cash charges added back. The manual warns that the traditional method "does not recognize growth in accounts receivable or inventory, a slow-down in accounts payable, capital expenditures, or additional borrowings", and says "The accrual conversion method is the preferred method because it is the most reliable".3

In the example, inventory up by about $692,000 is cash on the shop floor that EBITDA does not see. The line of credit, 94% drawn in August, is where that cash came from.

The ratios on most credit memos

SBA names "Current Ratio, Debt/Tangible Net Worth, Debt Service Coverage" as its benchmarks and asks lenders to compare them with industry trends.2 Peer comparisons usually come from statement studies built from banks' own uniform spreads, which is one more reason the RMA guide asks for spreads done the same way every time.1

Twelve spreading errors that make a borrower look stronger than it is

In shortSpreads fail in four places: in global cash flow, in the keying, in the package, and in the documents themselves. Most of the costly errors look right on the grid. Reading the spread over again catches little: in a controlled study, data checked by eye ended up with about as many errors as data never checked at all.

Why reading it over is weak protection

The study's subjects were undergraduates entering data sheets. Its lesson carries over to spreads: a person rereading a grid finds what looks wrong, and the costly spreading errors look right. Examiners see the result as unidentified exceptions, which the OCC says "are often the result of errors, miscalculations, breakdowns in the pre- or post-closing review process, or system failures".11

What a wrong spread risks: exam findings, loan losses, fraud and fair lending

In shortA spread that overstates coverage can approve a loan that should have been declined, or keep a weakening credit rated as a pass. Examiners look for weak financial analysis and unidentified exceptions. Fraud enters through documents the bank never ties to an independent source. And business credit still carries fair lending duties.

Commercial loans that stop paying

Net charge-offs on C&I loans ran at an annualized 0.57% across all insured institutions in the first half of 2026, and 0.50% at community banks.23 Those are averages across $2.5 trillion of loans. For one bank the loss that matters is the single credit approved on a ratio that was never real.

Credit conditions in 2026: standards steady, demand rising

More demand and steadier standards mean more packages to spread, and fewer of them declined on the first look. That puts more weight on the spread itself.

Exam findings, fraud and fair lending

The commercial lending market that spreads serve, in numbers

In shortUS commercial banks hold close to $3 trillion of commercial and industrial loans, and more than half of insured institutions are classed as commercial lenders. Most banks decide within ten business days, and only one in ten has credit scoring that can automate some underwriting28. One vendor's analysis of call-report data puts more than three in four institutions on a manual commercial lending process.

The FDIC's survey of about 1,300 banks found that "Three in four banks approve their typical loan within ten business days", that three in ten used financial technology in lending, and that "One in ten banks have a credit-scoring system that can partially or fully automate the underwriting of some non-credit-card lending".28 A four-entity spread that occupies a calendar week uses half of that typical decision window.

No regulator publishes how many banks still spread in Excel. One vendor's analysis of call-report and third-party data, published in a state bankers' association newsletter in July 2025, counted 6,352 institutions without "a commercial lending software solution" against 1,393 with one. It put the share still using a manual process at 77%, though its own two counts work out nearer 82%.31

The same analysis wrote that a majority use only Word and Excel in commercial lending.31 Read it as a vendor's count. Its claim that analysts spend "multiple days" on data entry rests, by its own footnote, on conversations with lenders.31

AI and automated financial spreading: what the 2026 model guidance covers, and what it leaves to your bank

In shortSoftware can read the documents, place each figure on a template line, do the arithmetic and draft the exceptions. Under SR 26-2, the arithmetic and fixed rules fall outside the definition of a model, and generative AI falls outside the guidance's scope. The controls then come from third-party guidance, safety-and-soundness standards and your own AI policy. The questions that matter are whether each figure can be checked against its page, and how often the tool misses what matters.

SR 26-2 defines a model as "a complex quantitative method, system, or approach that applies statistical, economic, or financial theories to process input data into quantitative estimates", and excludes "simple arithmetic calculations, such as those found within spreadsheets, as well as deterministic rule-based processes and software where there are no statistical, economic, or financial theories underpinning their design or use".5

A footnote puts generative AI, and AI that acts on its own, outside the guidance, while its principles still apply to traditional statistical models and to AI that neither generates content nor acts on its own. The same footnote says a bank's own risk management "should guide the determination of appropriate governance and controls" for tools it does not cover.5

The Fed expects it to be most relevant above $30 billion of assets; the OCC says it does not require community banks to validate models every year, and that non-compliance "will not result in supervisory criticism".5,8

What supervisors say about AI-read financial statements

The third-party guidance asks banks to assess a provider's ability to "perform the activity as expected" and keeps the responsibility with the bank.14 NIST's framework names four functions, GOVERN, MAP, MEASURE and MANAGE; for spreading, MEASURE means a test you design.36

Colorado's SB26-189 covers automated decisions about "financial or lending services" from January 1, 2027; it is aimed at consumers, and whether a commercial credit with personal guarantors falls under it is a question for counsel.10 The CFPB withdrew its circulars on adverse-action reasons for complex algorithms in May 2025; Regulation B's notice duties did not change.37,27

How you would know a tool works: plant errors and count

Accuracy has two parts. Transcription asks whether each figure on the spread is the figure on its page; a random sample measures it. Catching asks whether the spread surfaces a problem in the package, such as a double count or a missing page; such problems are rare, so they have to be planted. A small planted test proves less than it seems to.

A test of twelve errors, one of each type, is a demonstration. Twelve types with 30 to 60 plants each is a measurement. Keep clean copies too, and count false alarms on them, because a tool that flags everything catches every plant. We found no published independent test of any spreading vendor's accuracy, as of October 2026.

Judgment calls a spreading analyst keeps

• Judgment stays with people: whether rent to an affiliate is an expense to eliminate, whether a 19-month-old personal statement can be relied on, whether a margin squeeze is temporary.
• A package without its last K-1 still waits on the borrower.
• The memo, the committee and the closing take the time they take. "Nearly three in four banks have at most three levels of approval".28
• The responsibility for the spread stays with the bank that signs it.14

How credit teams spread today: who does what, and where the days go

In shortA relationship manager gathers the package, an analyst sorts, keys, ties out and combines it, a senior underwriter or credit officer signs, and a committee decides. Most of the analyst's hours go to keying and tying out. The work peaks in the autumn, when extended returns arrive and annual reviews pile up.

Relationship manager to loan review: who handles a spread

Three people handle a spread before committee

Keying takes about two-thirds of a four-entity spread

Few analysts get three uninterrupted days. A spread that needs 24 hours of attention can occupy a calendar week between calls, other packages and a memo due for another committee.

The calendar adds its own pressure: S corporation and partnership returns for calendar-year companies are due March 15, and a six-month extension moves them to September 15; extended personal returns run to October 15. Annual reviews bunch up in the autumn.16,38,39,40 Most business credit lines come back for review every year: among firms with under $50 million of assets, "three-quarters of credit lines have maturity of 1 year or less at origination".41

Analyst workbook, spreading service or AI extraction tool

The relationship knowledge in a spread lives with whoever built it. The Fed's Berger and Udell made the general point in 2001: "Because the loan officer is the repository of this soft information, agency problems are created throughout the organization".42 A mapping decision kept in one analyst's head is a small version of the same problem.

The Arnevik package: 38 documents spread, five flagged lines resolved, and the senior underwriter's co-signature

In shortThis is the example from the top of the page, inside the product. Six agents sort, read and spread the 38 documents minutes after the last K-1 arrives. Pallavi Natesan, the commercial credit analyst, works down five flagged lines with each page beside its figure, decides the five exceptions and signs. Walter Brandt, the senior underwriter, co-signs because the relationship is over her $1,000,000 limit.

1

Nine packages, ranked by when the memo is due

Monday morning, Pallavi's queue holds nine packages: 27 entities to spread or review, four credit memos due for Thursday's committee, nine open exceptions. Arnevik sits at the top: 38 of 38 documents received Friday at 4:40 p.m., four entities, memo due Wednesday.

The queue also shows what is waiting on borrowers: a trucking company's missing 2025 K-1, a farm equipment dealer's interim that is 165 days old.

2

The package was spread before anyone opened it

When the last K-1s arrived on Friday, the package sorter matched 38 documents to four entities and three tax years, 412 pages, none missing. The tax return reader read 12 returns and 12 K-1s and tied them. The statement spreader spread three reviewed statements and the August interim. The run finished in 6 minutes 48 seconds.

The package screen shows who is in the credit and how they are owned, drawn from the K-1s, and five exceptions with their pages: Schedule E income is not cash, distributions counted once, an interim balance sheet $50,000 out, a margin slip, an old personal statement.

3

Every figure opens its page

Pallavi opens the operating company's spread: three fiscal years in the bank's chart of accounts, every cell carrying a pin, amber for the tax return, blue for the reviewed statement, a ring for a calculated line. She hovers 2025 officers' compensation, $620,000. Page 1 of the Form 1120-S slides in with line 7 boxed.

The tax return reader has tied it to page 4 of the reviewed statement and to Form 1125-E: Erik Arnevik $420,000 plus Marcus Delacroix $200,000, the same as both W-2s. She confirms it.

4

Global cash flow, counted once

The global cash flow builder combined the four entities under the bank's combined entity method: 1.47x for fiscal 2025, against the 1.84x of last year's spreadsheet, which added $591,000 of the owners' Schedule E income on top. The screen shows the two methods side by side, with the Schedule E line and the K-1 box behind the difference.

On the trailing twelve months, global coverage holds at 1.31x but business coverage is 1.16x, under the 1.20x policy. A stress panel recomputes both as margin or revenue moves.

5

Five exceptions decided, then two signatures

Pallavi accepts the two agent adjustments, Schedule E and distributions. She notes the other three for the credit memo, and asks for the September interim, the supplier's repricing letter, the terms of the $50,000 owner advance and Marcus Delacroix's updated personal statement. Then she signs: "Spread complete and tied to source".

Walter Brandt co-signs: "Approved for the credit memo". The spread goes to Thursday's committee with every figure linked to its page.

The solution's design target is under three hours of analyst time for a signed four-entity spread like this one. It is a target, to be measured on your own packages.

Eleven more screens, from a re-mapped line to the bank's own credit policy

In shortOne package is the story. These screens show the rest: the spread itself, a line moved to the right account and learned for next year, a line split three ways, the interim and trailing twelve months, the ratios against peers, the co-signed record, the relationship's history and the bank's own policy.

Select any screen to open it larger. The bank, borrowers, people and figures are fictional; peer figures are sample data.

Keying and tie-outs by agent, with every spread signed by the analyst and co-signed above $1,000,000 of exposure

In shortEach agent has one job and limits it does not cross. None of them signs a spread, sends a request to a borrower, or plugs a statement to make it tie. The credit analyst signs every spread; above $1,000,000 of exposure a senior underwriter co-signs. Every adjustment carries its reason, and every learned rule carries a name.

Package sorter to exception writer: the six spreading agents

The analyst signs every spread; a senior underwriter co-signs large ones

Signatures record the name, the decision, what it means and the time. Agent steps and human decisions are kept apart on the activity record, every spread version is kept, and nothing goes to a borrower until the analyst sends it.

The analyst, underwriter and Arnevik CFO in the screens

Borrower packages spread into the bank's own template, with page-linked figures, global cash flow, ratios and two signatures

In shortIt reads the documents borrowers already send, writes into the bank's own template, and leaves behind a multi-year spread for every entity with each figure linked to its page, global cash flow, ratios, exceptions with drafted requests, and two signatures. The bank's systems of record stay where they are.

Systems a spreading rollout connects to

For each, check whether it has an API, a file feed, or only screens, and who owns it.

Your template, origination system and earlier spreads stay

Page-linked, counted once, signed by an analyst

What a hand-built spread costs you in hours, waiting days and flattered ratios

In shortStart from your own costs. No regulator or study publishes the time a spread takes or how often spreads are wrong, so the numbers that matter are yours. A few published figures frame them: what an analyst hour costs, how often commercial loans go bad, and how fast other banks decide.

Timing five spreads to find your own baseline

Credits you should have declined, and borrowers who left

What faster spreads return when you count analyst hours against the tool's yearly cost

In shortPut in your own numbers. The calculator starts from a published analyst wage and asks you for the rest: how many spreads you do, the hours they take now, the hours you expect after, and what the tool and your testing cost a year. Every default marked illustrative is an assumption to replace.

Keying, mapping and double counts: where spreading saves

Spreads a year, hours keyed and your hourly cost

A bank that spreads 200 owner-managed relationships a year at renewal and adds 60 new ones clears the first two lines and sits just below the third. A bank that spreads 40 audited single companies a year probably is not, and should say so before buying anything.

How ready your spreading desk is: volume, template discipline, documents, errors on file and sign-off

In shortPick the answer closest to your team today. Eight questions score 0 to 2; the last is for reflection. The meter adds up your score as you go, and each question says why it matters and what a strong answer sounds like.

A low score means settling one spreading template first

What to ask spreading software or a spreading service about pass-through income, evidence and your template

In shortAsk every vendor the same questions, us included. The answers show how the spread is built: what it reads, how it counts pass-through income, what evidence you keep, whose template and rules apply, and how you would measure it on your own packages.

We found no published independent test of any spreading vendor's accuracy, as of October 2026. A planted-error test on your own packages is the fairest comparison, and the same test measures your current process.

Test spreading software on your own packages with planted errors, then decide on three numbers

In shortMeasure first, then test on your own packages with errors you planted, then decide with three numbers. None of these steps commits you to a vendor.

1. Time five spreads, one single-entity and four multi-entity, by step. That is your baseline.
2. Pull last year's evidence: loan review and exam comments on financial analysis, and any credit whose global coverage changed on review.
3. Gather 40 recent packages with your template and policy, and an approval path to share them. Copy ten and plant known errors in them.
4. Run in parallel on live packages for five weeks, with your analysts signing every spread.
5. Decide with three numbers: catch rate by error type, false alarms on clean packages, and analyst minutes to a signed spread.

Catch rate, false alarms and traceability per planted error

Financial spreading questions answered: global cash flow, K-1 versus Schedule E, SBA coverage and statement levels

In shortThe questions credit teams and buyers search for most, each answered on its own.