ABL field exams, explained.

A lender's quarterly visit to check the collateral behind a revolving loan

Where the field exam sits in an asset-based loan, in five steps

1. The line is set upA first exam and an inventory appraisal before closing; the loan agreement defines eligible receivables and inventory, advance rates and reserves.
2. The borrower certifiesWeekly or monthly certificates with agings and inventory reports; the lender advances against them.
3. The lender monitorsCash through a lockbox, availability tracked between exams, trends watched.
4. A field exam tests it allUsually quarterly for weaker credits and once or twice a year for strong ones. Roll-forward, ineligibles, dilution, samples, counts, a report to credit.
5. Credit acts on the findingsEligibility and reserves change, the borrower is told, and the next exam is scheduled by risk.

Step 4 is the subject of this guide. The day-to-day checking of each certificate and the monitoring between exams have their own guides.

Four facts about field exams that surprise newcomers

Four terms every exam report uses

Fourteen terms are defined in the one-page summary.

What a field exam tests, from the certificate to the payroll taxes

In shortAn exam tests the whole working capital cycle: sales become receivables, receivables become cash, cash buys inventory, and payables and taxes compete for the same cash. Supervisors list the same core tests: the certificate reconciled to the books, concentrations, trends, control of cash, taxes paid, verification with customers, credit memos and inventory. Each test answers one question about whether the collateral is real and collectible.

The Federal Reserve's manual gives the scope of an audit as one list. It starts with "verification that the information on the borrowing-base certificate reconciles to the borrower's books", then "review of concentrations of accounts", "review of trends in accounts receivable, accounts payable, inventory, sales, and costs of goods sold" and "review of the control of cash proceeds".

The list goes on with a check of the general ledger, "verification that all taxes, especially sales and payroll, are paid timely" and "review of compliance with the loan agreement".1 The same section asks for a program of "direct confirmation" of receivables, "before the initial lending arrangement and periodically thereafter", on a positive basis.1

The FDIC's examination module, revised in October 2025, tells its examiners to determine whether "periodic on-site audits of borrowers are performed" and, if so, how often, and to check that they look at shipping documents, inventory invoices, turnover, "Direct verification of AR and AP", concentrations, "Sales dilution", the borrower's collection procedures, affiliate receivables and over-advances.4 Practitioners describe the same scope as "the full working capital cycle—cash, accounts receivable (AR), inventory, and accounts payable (AP)", plus payroll tax testing.7

Four questions every exam answers

Which supervisory texts expect field exams, and which of them 2026 left untouched

In shortNo federal rule fixes how a lender runs a field exam. Bank supervisors expect an exam program in the lender's own policy and judge it against their manuals: the Federal Reserve's section 2160 (May 1996), the OCC's Asset-Based Lending booklet (2014) and the FDIC's examination module (October 2025). In 2026 the model risk guidance and the OCC's general lending booklet were replaced; the asset-based lending texts were not.

The auditors' standard does not bind a lender. It is the clearest recent statement of how to verify a receivable: the person verifying "should select the items to be confirmed, send confirmation requests, and receive confirmation responses", and negative requests, the kind that ask a customer to reply only if something is wrong, give evidence "only when combined with other substantive procedures".8 A field examiner who lets the borrower mail the verifications, or forward the replies, has given up the point of the test.

How often a borrower is examined, and who pays for each visit

In shortSupervisors say quarterly for most borrowers and more often when risk rises. Loan agreements turn that into a count the borrower pays for: often one exam a year while availability is comfortable, two when it falls below a threshold, and as many as the lender wants after a default. The borrower pays a daily rate per examiner plus expenses.

The OCC's booklet, in its section on field audits, says a field audit should come before a new account is booked and regularly after that, often quarterly but more often if the risk calls for it, with weekly or even daily audits for a workout borrower.5 The Fed's manual says audits are "usually performed quarterly" for satisfactory borrowers.1 The FDIC asks its examiners only to "determine the frequency" and judge it.4

Contracts filed with the SEC show how lenders write it down. A 2024 amendment to one manufacturer's revolving credit agreement lets the lender bill "one field examination per calendar year", one more if excess availability has been at or below 25% of the maximum revolver amount, and "each other field examination ordered during the existence of an Event of Default".14 A 2023 bank agreement with another manufacturer allows one exam in any 12 months while excess availability stays at or above 15% of the maximum credit, and two if it ever falls below.3

What an examiner-day costs

Why does the borrower pay? The OCC's booklet notes that smaller banks may use independent staff lenders or outsource the work, and that either way the borrower generally bears the cost (section on field audits).5 An exam that runs four examiner-days on site and six off site costs about $10,000 to $11,000 at those rates, before travel. That figure is our arithmetic on the published rates. Nobody publishes a survey of exam fees or exam lengths.

The files an examiner asks for, and the workpapers built from them

In shortBefore the visit the borrower sends a data download: about a dozen files that cover the period since the last exam. The examiner turns them into workpapers, one per test, each tying a figure to a source. A good set of workpapers lets someone who was not there follow every number in the report back to the borrower's own records.

The examiner's discipline is the same one an auditor follows: every figure on a workpaper names the file and row it came from, and a difference that cannot be explained stays on the workpaper as a difference. Practitioners point out that the data rarely arrives clean: examiners "clean and standardize data" from PDFs, inconsistent file layouts and complex corporate structures before any test can run.7

The example's download was 14 files, received Friday, October 9, 2026 at 4:48 p.m., for fieldwork on Tuesday, October 13 to Thursday, October 15.

Rolling receivables forward to the dollar, and what the general ledger difference means

In shortStart from the aging the last exam tested, add the sales since, subtract the cash, credits and write-offs, and you should land exactly on today's aging. Then tie that aging to the general ledger control account and to line 1 of the certificate. A roll-forward that ties proves the aging is complete. Whether each invoice is real is a question for the samples that follow.

In the example, the June 30 aging was $18,412,560. Over the quarter Tamsin billed $31,846,200, collected $30,118,940, issued $1,431,880 of credit memos and wrote off $62,300. The result is $18,645,640, which equals the September 30 aging and line 1 of the September certificate.

The general ledger control account read $18,589,140, $56,500 lower. The examiner does not force it. The difference turns out to be cash received on September 30, posted to the general ledger that day and applied to invoices on October 1: unapplied cash, which the certificate had already deducted. That explanation goes on the workpaper with the receipt.

The same roll-forward, month by month, gives the examiner the monthly sales, cash and credit figures that dilution and turnover are built on. In the example, September's credits were $757,400 on $10,552,300 of sales, 7.2% as reported, more than double July's 2.9%. That jump is the thread the next sections pull.

How one rebilled customer becomes wholly ineligible under the 50 percent rule

In shortLoan agreements exclude receivables that are too old, disputed, owed by affiliates or foreign customers, or billed before shipment. The examiner recomputes each class from invoice detail. Two rules do most of the work: age is measured from the original invoice date, and once enough of a customer's balance is past due, all of it is excluded. A credit-and-rebill is designed to beat both.

The Fed's manual explains why delinquency is measured from the invoice date: an agreement that says "90 days from the invoice date when payment is required in 30 days" stops a borrower from keeping late accounts eligible "by giving dated terms (extending payment days)". Under the "50 percent rule", if a customer owes ten invoices and six are delinquent, "all ten would be considered ineligible", including the four that are not yet late.1

The same section lists contra accounts, affiliate accounts, concentrations ("no more than 20 percent of the receivables assigned should be from one customer" as "A common benchmark"), bill-and-hold sales and progress billings.1

The Holloway rebill, worked through

On September 28, Tamsin credited 14 Holloway Packaging invoices first dated June 15 to July 1 and reissued them the same day for the same $186,240. As reported, those invoices were two days old. Measured from their first dates they were 91 to 107 days old at September 30, past the agreement's 90 days from invoice. That put $186,240 of Holloway's $342,600 balance, 54%, past due: above the 50% cross-aging limit, so all $342,600 is ineligible. At an 85% advance rate that is $291,210 less availability.

Across the book, recomputing every class from invoice detail raised receivable ineligibles from the certified $1,912,400 to $2,354,960: Holloway's $342,600, $61,360 billed before shipment, and $38,600 a customer disputed in its reply. Receivables availability fell from $14,223,254 to $13,847,078.

Measuring dilution once the rebills are taken out

In shortDilution is the share of sales that never turns into cash because of returns, allowances, pricing credits and the like. It decides how much cushion the advance rate needs. Measured carelessly it is wrong both ways: rebills inflate it, and returns that are authorized but not yet credited hide the next month's credits. The examiner reads the credit memos one reason at a time.

The Fed's manual says the advance rate must leave "a sufficient margin for the continual costs of collection and absorption of dilution", and that "A lender must know how dilution is occurring in each receivable portfolio to measure it continually". Returns, bad debts, product liability and warranty claims are the negative causes; discounts and allowances can be benign.1 The OCC's booklet adds that dilution is usually expected to be 5 percent or less, and that a rise can signal falling product or service quality.5

In the example, Tamsin's credit memos for the quarter were $1,431,880 on $31,846,200 of sales: 4.5%. But $186,240 of those credits were the Holloway rebill, which cancels an invoice and reissues it. Taken out of both credits and sales, true dilution is 3.9% for the quarter. September alone is 5.5%, against 2.8% to 3.6% in each of the eleven months before it and a 5% trigger in the loan agreement.

Credits by reason tell you what to fix

The returns log matters as much as the credit memo register. Nine returns from one customer, worth $214,800, were authorized in September but not yet credited at cut-off. Those receivables are on the aging at full value and will shrink within weeks. The examiner recommends a reserve for the full amount, so availability already reflects credits the borrower has promised but not yet issued.

Sampling invoices, receipts and customer replies, and what a clean sample proves

In shortThree tests check that receivables exist. The sales test traces sampled invoices to bills of lading and proof of delivery. The cash test traces receipts from the bank statement to the aging. Verification asks customers directly what they owe. Each sample should record how it was drawn, so a reviewer can tell whether a clean result means anything.

Sales to shipping: catching pre-billing

The Second Circuit's Carboni case shows the test working. Cableco, a Connecticut distributor of wire and cable products, pre-billed invoices for goods not yet shipped from September 1993 to January 1994 and listed them as eligible on its certificates. The bank found the scheme at a February 1994 audit: its examiner found about $547,000 of advances not supported by eligible collateral and testified that some accounts were pre-billed "by as much as fifty-eight days". The bank's net loss was about $674,230.16

In the example, the sample was every September invoice over $100,000 (six of them) plus 34 drawn at random from the rest with a recorded seed: 40 invoices, $2,936,400, 27.8% of September's sales. Thirty-seven matched the shipping log. Two invoices to one customer, dated September 29 and 30 for $61,360, left the dock on October 2, so they were not receivables at cut-off. The third, a $55,060 order drop-shipped by Tamsin's supplier, had no record in Tamsin's own log; on site the carrier's proof of delivery dated September 28 cleared it.

What a clean sample can and cannot tell you

A sample with no exceptions does not prove there are none. With zero exceptions in n items, the 95% upper bound on the exception rate is roughly 3 ÷ n, a result statisticians call the rule of three.17 Forty clean invoices still allow an exception rate of up to about 7%. That is why examiners also stratify, taking every large item. Where the data allows it, a test of every invoice against the shipping log says more than any sample.

Cash: following the money, and who paid it

The cash test samples receipts and traces them from the bank statement through the cash receipts journal to the invoices they paid. It catches misapplied cash, cash that never reached the lender's account, and, if the examiner looks at who paid, cash that came from the wrong place.

In the Allou case, the indictment says the company booked hundreds of millions of dollars of fictitious sales, then recorded transfers from family-affiliated companies "as payments for the fictitious sales", while "The Lenders' auditors conducted audits of Allou on a quarterly basis".18 The SEC's complaint put the falsified sales at about $220 million in the fifteen months to March 2003.19 A receipt that matches an invoice proves little if the payer is not the customer.

Verification: asking the customer

The Fed's manual asks for "a program of direct confirmation", on "a positive basis".1 In the example, the lender sent verifications to the twelve largest customers and eight at random, $9,701,800 or 52.0% of receivables, with replies addressed to the lender. Sixteen confirmed, three were cleared by cash received after cut-off, and one customer disputed $38,600 for defective couplings. The disputed amount became ineligible.

Counting the shelves: test counts, cost tests and slow stock

In shortThe inventory work ties the perpetual records to the general ledger, ages every item by its last sale, counts a sample of shelves and checks costs against supplier invoices. The count tests whether the records match the floor; the aging tests whether the certificate excludes what the loan agreement calls slow or obsolete. Most inventory findings are about age; missing stock is rarer.

Record errors are common even without fraud. In a study of nearly 370,000 inventory records at 37 stores of one retailer, 65% did not match what was on the shelf, and the researchers found that auditing practices reduced the errors.20 A distributor's warehouse differs from a store, and the lesson still carries over: a perpetual record is a claim to be tested, and a count of well-chosen items is the test.

The count

Examiners count in both directions: from the sheet to the floor, which tests that recorded items exist, and from the floor to the sheet, which tests that the records are complete. Counts are weighted to value. In the example, the count covered 60 SKUs, 40 chosen from the 500 most valuable and 20 at random, $612,480 at book value. Fifty-seven agreed; three differed by a net $9,840, 1.6% of counted value, inside the 2% tolerance in the loan agreement. A cost test of 25 SKUs against supplier invoices found every one within 5%.

Video counts became common during the pandemic. One firm described counting from the floor to the records with warehouse staff opening boxes on camera, and still said "we would make visits if inventory was involved".21

Slow stock the certificate left in

Tamsin's perpetual held 18,412 SKUs, $14,280,600 at cost, tying to the general ledger. Aged by last sale, $1,064,300 had not sold in 12 months. The certificate excluded only $612,000, because the borrower counted only SKUs with no sale in 18 months, while the loan agreement says 12. The difference, $452,300, is ineligible; at a 50% advance rate, availability falls by $226,150.

What court and agency records show field exams catching, and being fooled by

In shortThe public record shows both. A 1994 bank audit caught pre-billing within months. Quarterly audits at a health and beauty distributor were defeated for years by fabricated documents and cash routed through affiliates. The 2025 and 2026 losses tied to an auto parts maker, a subprime auto lender and a materials distributor turned on receivables or collateral that prosecutors or the lender say were fake, misrepresented or pledged twice. The lesson for an exam is in the mechanics each scheme relied on.

Practitioners list the same patterns: "pre-billing receivables before shipment", "refreshing the A/R aging by credits and rebills to current receivables", and collusion with "phony vendors, phony customers", dishonest truckers or warehouse staff.6 Supervisory texts add bill-and-hold, progress billings and affiliate accounts.1,4 The 2025 and 2026 cases show why the cost of a missed scheme is measured in the loan itself: a bank expected a non-cash impairment of $170 million to $200 million on a loan of about $200 million, and another expected to charge off $22.0 million of $23.6 million.23,27

More asset-based deals and leaner teams: the numbers behind exam workloads in 2026

In shortThe lenders that answer the Secured Finance Network's quarterly survey in every quarter held about $320 billion of committed asset-based lines at the end of 2024. That figure covers a fixed panel of respondents, which is part of the market. New deals rebounded in 2026, non-bank lenders grew faster than banks, and the lenders' own head counts fell in 2023 and 2024. More borrowers per examiner is the direction of travel.

The 2026 surveys show activity picking up. In the fourth quarter of 2025, non-bank lenders' outstandings rose 12.6% in the quarter and their commitments 5.5%, while banks' commitments rose 0.5% and their outstandings fell 5.1% on seasonal paydowns.29 In the second quarter of 2026, 36 lenders reported new commitments to new clients up 58.7% at banks and 60.9% at non-banks on the quarter.30

Credit quality at banks was good by long-run standards: criticized and classified loans were 10% of outstandings, non-accruals 0.78% against a 30-year average of 0.92%, and gross write-offs 0.09% against 0.41%; at non-banks, criticized loans and non-accruals rose.30 Meanwhile total employees at the reporting lenders fell 2.5% from 2022 to 2023 and 4.2% from 2023 to 2024.28 No public survey counts field examiners; the employee figure is the closest public proxy, and it points one way.

Who examines, who signs, and where the weeks between fieldwork and report go

In shortA field examiner, often one of a small team covering dozens of borrowers, builds the workpapers, visits and drafts the report. A field exam manager schedules, reviews and signs. A credit officer approves changes to eligibility and reserves; the portfolio manager tells the borrower. The examiner's days go mostly to preparing and writing up, and the report waits for review.

Field examiner, exam manager, credit officer and the borrower's controller

The Fed's manual asks that the people who perform audits "be independent of the credit function".1

Four people between the exam request and credit's action

Where an examiner's hours go

The same practitioner argues that the worksheets have crowded out the judgment, and that "you do not need 50 or 75 to achieve an adequate report".6 A 2022 round table listed what a good report shows: key findings, test results and exceptions, and proposed eligible and ineligible collateral with the calculations behind them; it also warned that "An overly aggressive timeline may be unrealistic".31 Practitioners also stress data security: examiners handle "highly sensitive, non-public information" that "must be protected from cyber threats and unauthorized access".7

Three ways lenders run exams

Remote work stayed after 2020. One firm had done desktop exams "for more than five years as a necessity to save costs" by 2021; another examiner expected exams to become "partially remote and partially at the borrower", and saw few billable hours saved.32,21 The analysis can travel; the count, the warehouse walk and the controller's answers mostly cannot.

The Tamsin exam: 14 files on Friday, three days on site, a report signed the next morning

In shortThis is the example from the top of the page, inside the product. The exam program runs the moment the borrower's 14 files arrive: 32 steps finished before the visit, one stop where a person must look, and seven steps left for the site days. Owen Castellano, the senior field examiner, confirms five findings on site; Denise Haugland, the field exam manager, signs the next morning; the credit officer approves the changes to the borrowing base.

1

The exam schedule, with Tamsin ready to prepare

Monday, October 12. Owen's exam list shows 31 of 38 borrowers due this quarter, one overdue, and a median of 3 days from fieldwork to report against 19 days in 2025. Tamsin Industrial Supply is at the top: quarterly, cut-off September 30, fieldwork Tuesday to Thursday, report due Thursday, October 22, five business days after fieldwork.

Gail Prentiss, Tamsin's controller, uploaded the 14 files on Friday at 4:48 p.m.

2

The program runs, and stops on the sales test

Owen runs the exam program: 39 steps in eight areas, from receivables to the borrowing base. Twelve steps in, it stops. Of September 40 invoices sampled, 37 agree with the shipping log and three have no shipment at billing. Two Pembury Conveyor invoices, billed September 29 and 30, left the dock on October 2. A $55,060 Galloran Truck Bodies order was drop-shipped by the supplier, so only the carrier's proof of delivery can clear it.

The held card shows the sample's method beside the three invoices: every invoice over $100,000 plus 34 at random, seed 2610-05.

3

Thirty-two steps done before the visit

Owen holds the three invoices for the shipping office on Tuesday at 10:30 a.m. and the program finishes in 1 minute 23 seconds of agent time. Every step that needs no eyes is done: receivables tied to the dollar, the Holloway rebill found, true dilution measured, inventory aged, availability recomputed.

The seven steps that need a person become his site days: the opening meeting, the bills of lading, the approvals on credit memos over $10,000, the floor count, the slow bays, the bank statements and the exit meeting.

4

Eleven workpapers from fourteen files

The exam overview shows what each agent did and how long it took: 41,862 sales lines, 9,114 receipts, 1,206 credit memos and 18,412 SKUs read; four samples drawn; five findings raised. Availability as certified is $19,693,554. After the findings it is $817,126 lower, and excess availability of $1,926,428 sits below the $2,500,000 springing level.

5

Every finding confirmed on site

Thursday afternoon, Owen confirms each finding against what he saw. Gail Prentiss confirmed the June invoices were reissued at Holloway's request with terms still running from June. Both Pembury bills of lading are signed by the carrier on October 2. Seven of the nine Ardent Robotics returns are in the receiving bay. The Galloran drop shipment was cleared by the carrier's proof of delivery dated September 28.

Unticking a finding drops it, but it stays on the workpaper with his reason, and the manager sees it.

6

Signed off, then to credit

Owen submits the report on Thursday, October 15 at 5:20 p.m. Denise Haugland signs it off on Friday, October 16 at 8:42 a.m., one day after fieldwork. Owen then sends the five changes to the borrowing base to Ruth Okafor, the senior credit officer, with the workpaper behind each one. Marcus Tellwood, the portfolio manager, tells the borrower after she approves.

The solution's design target is a signed report one to two days after fieldwork, with every finding tied to its workpaper. It is a target, to be measured on your own exams. The example's dates follow the sample data; October 12, 2026 is Columbus Day, and nothing in the example depends on a bank or the Federal Reserve being open that day.

The workpapers behind the report, screen by screen: ties, ineligibles, dilution, samples, stock and trends

In shortOne exam is the story. These screens show the workpapers behind it: the roll-forward and the re-aged aging, every ineligible recomputed, dilution with the rebill taken out, the three tests of existence, the inventory, the report's borrowing base impact, the site plan, the borrower against its history and the book, and the rules the exam is judged by.

Select any screen to open it larger. The lender, borrowers, customers, people and figures are fictional sample data.

What each agent prepares for the field examiner, and who confirms, signs and approves

In shortEach agent has one job and limits it does not cross. None of them edits the borrower's data, calls a pattern fraud, issues a report, changes a borrowing base or contacts the borrower. The field examiner owns every conclusion and confirms it on site, the field exam manager signs every report, and a credit officer approves each change to eligibility or reserves.

Six agents read the download and draft the workpapers and report

Examiner confirms, manager signs, credit officer acts

Every agent step and every human decision is recorded with who, what and when. Edits to a report make a new version, and changes to the exam rules are versioned too.

Owen Castellano's exam team and Tamsin's controller

From the borrower's data download to cited workpapers and a signed field exam report

In shortThe solution reads the files borrowers already send, runs your exam program under your loan agreements' definitions, and leaves behind workpapers with every figure tied to its file and row, samples with their method on record, a draft report that cites every test, and the signatures it needs. Your loan and collateral systems stay where they are.

Borrower portal, collateral system and lien searches to connect

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

Your exam program and loan agreements set the checklist

Three rules every exam follows

What weeks of workpapers and a late report cost a lender

In shortStart from your own numbers. Nobody publishes how long exams take or how often they change availability, so your exam log is the evidence. Three published figures frame it: what an examiner-day is billed at, how long review can take, and what a missed receivables fraud costs when it surfaces.

Log your examiners' hours by exam phase

Advances on a rebill, fee-weary borrowers and departing examiners

Your return on field exam automation, priced in examiner days and review days

In shortPut in your own numbers. The calculator starts from a published wage for auditors and asks for the rest: exams a year, examiner and manager days per exam now and after, and what the tool and your testing cost a year. Every default marked illustrative is an assumption to replace.

Workpapers built before the visit, reports days after fieldwork

Run the numbers on your own exam schedule

A team of seven examiners doing 14 exams each a year runs about 100 exams and clears the first two lines comfortably. A lender with ten large, clean borrowers examined once a year does not, and should say so before buying anything.

A field exam readiness score: your baseline, exam program, data downloads and governance

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.

From three timed exams to a new standard of preparation

Seventeen questions on rebills, samples, workpapers and proof for field exam software or an exam firm

In shortAsk every vendor the same questions, us included. The answers show what the tool reads, which tests it runs on every item, how it treats rebills and returns, what evidence you keep, whose rules apply, and how you would measure it on your own exams.

We found no published independent test of any field exam tool's accuracy, as of October 2026. A test on your own past exams, with planted problems, is the fairest comparison, and the same test measures your current process.

A ten-week field exam pilot on past downloads whose findings you already know

In shortMeasure first, then test on past downloads with known findings and some planted ones, then run live exams in parallel, then decide with three numbers. None of these steps commits you to a vendor.

1. Time three exams by phase, and pull last year's fieldwork and sign-off dates. That is your baseline.
2. Write down the program: tests, sample rules, report format, and the agreement terms of the borrowers in the pilot.
3. Gather 20 past downloads with their final reports. Copy five and plant problems: a rebill, a pre-billed invoice, an unrecorded return, a slow SKU left in.
4. Run four live exams in parallel with your examiners doing their usual work and signing the reports.
5. Decide with three numbers: findings caught by type, false alarms on clean copies, and examiner and manager days.

Findings caught by type, false alarms and days to a signed report

Field exam frequency, fees, roll-forwards, rebills and remote exams: common questions

In shortThe questions lenders, examiners and borrowers search for most, each answered on its own.