To calculate fraud loss from bank records, fix the period from the first to the last fraudulent transaction, list every payment in it that fits the scheme, remove the ones backed by real goods or services, subtract any recoveries, and show the result in confirmed and probable tiers. Every line in the total should trace to a statement page.
What goes into a fraud loss calculation from bank records?
A fraud loss calculation starts with the gross amount that left the business through the scheme, then removes anything legitimate and anything recovered. The result is net loss. Bank records supply the gross figure, because they show each payment's date, amount and payee; invoices, approvals and receipts decide which payments were legitimate.
The building blocks are the same in most cases:
- Gross payments in scope: every transaction that fits the scheme's pattern, such as all payments to a fake vendor or all checks the subject wrote to themselves.
- Less legitimate items: payments in that list with proof the business got value.
- Less recoveries: money returned, seized, repaid or paid by an insurer.
- Plus related costs, where the audience allows them, such as bank fees caused by the scheme or the cost of the investigation.
Decide early whether you are calculating one scheme or several. A clerk who ran a fake vendor and also padded expense claims has two schemes with different evidence. Give each its own schedule and add them only in the summary.
Keep these as separate lines, never one blended total. An insurer, a judge or a board member will want to see each step, and they may accept some lines while rejecting others.
Setting the fraud loss period: first and last transactions
The fraud loss period runs from the first transaction that fits the scheme to the last one. Set both ends from the evidence, not from when the suspicion arose. Look back further than seems necessary, since schemes often start small and grow, and the first payment is easy to miss.
To find the start, pull records at least a year before the earliest known fraudulent item and look for the same payee, account or pattern. To find the end, run through the date of discovery and a little past it, in case payments were already in motion.
Items at the cut-off
Items near either end of the period need care. A fraudulent check written before discovery may not clear until after it. If you stopped it in time, there is no loss; if it cleared, include it on the date it cleared. On the other side, a deposit the subject made to cover a shortfall may sit in transit at the end of the period. The rules for treating these timing items on a reconciliation are set out in outstanding checks and deposits in transit, explained, and the same logic decides whether they belong in the loss.
Separating theft from legitimate items in the fraud loss schedule
Separating theft from legitimate items means classifying every payment in the gross list as fraudulent, legitimate or uncertain, with the evidence for each call. A fake vendor's payments are usually all fraudulent. A real vendor paid inflated prices is harder: only the excess over a fair price is loss. Document the rule you used for each class.
A worked fraud loss schedule
The figures below are invented. They show payments to one vendor over 26 months.
| Line | Description | Confirmed | Probable | Excluded |
|---|---|---|---|---|
| 1 | Payments to "Lakeview Services," 2024-05 to 2026-06 (58 payments) | |||
| 2 | No invoice, no delivery record (41 payments) | 142,600.00 | ||
| 3 | Invoice exists, no evidence of work (12 payments) | 38,400.00 | ||
| 4 | Invoice and signed delivery note (5 payments) | 14,250.00 | ||
| 5 | Gross loss before recoveries | 142,600.00 | 38,400.00 | |
| 6 | Less: subject's repayment, 2026-07-20 | -10,000.00 | ||
| 7 | Net loss | 132,600.00 | 38,400.00 |
Lines 2 to 4 together account for all 58 payments, which is the check that nothing was dropped. The confirmed column is the figure you can defend line by line. The probable column needs a judgment call, and saying so openly is stronger than folding it into one total.
Uncertain items
For each uncertain payment, say what evidence would move it. "Would move to confirmed if the supplier confirms no work was done" gives the reader a path and gives you a task.
Recoveries and repayments that reduce the fraud loss
Recoveries are amounts the business has got back: repayments by the subject, funds frozen and returned by a bank, chargebacks, and insurance payments. Each one reduces the net loss, and each should be listed with its date and source. Whether a recovery counts depends on whether the money actually arrived, not on a promise to pay.
Common recovery lines:
- Repayments the subject made, sometimes under the guise of "returning a loan."
- Funds a bank recalled or froze after the business reported the fraud.
- Chargebacks on card payments the scheme used.
- Insurance proceeds, which you show separately because an insurer may reduce its payment for other recoveries.
Watch for disguised repayments. A subject who deposits money back into the account before discovery may be covering a shortfall, not returning stolen funds, and a schedule that ignores that deposit overstates the loss. Search the statements for deposits from the subject or related parties and treat each one as a possible offset.
Should interest be added to a fraud loss figure?
Interest on a fraud loss belongs in the figure only when the audience allows it. A court may award interest on a judgment under rules that differ by jurisdiction; an insurance policy usually defines what it covers and often excludes lost interest. Calculate the base loss first.
Add interest as a separate, clearly labelled line only if counsel says it applies.
If interest is relevant, the calculation needs a rate, a start date for each payment and an end date. Applying one rate to the whole total from the first payment overstates it, because most of the money left later. Compute interest payment by payment, or by month, from the date each amount left.
Lost opportunity, such as profit the business might have made on the money, is a separate and harder claim. It needs evidence that the money would have been used profitably, and it is often disputed. Keep it out of the main schedule unless counsel asks for it.
Presenting the fraud loss figure to an insurer, court or board
Present the fraud loss figure as a short summary backed by schedules: the net loss in confirmed and probable tiers, the period, the method and a list of what was not tested. Each audience reads it differently. An insurer checks it against policy terms, a court tests it through opposing experts, and a board wants to know what failed.
What each audience usually expects:
- Insurer: the claim form, a schedule per scheme, copies of the supporting statements and evidence that the loss falls in the policy period.
- Court: an expert report setting out method, assumptions and sources, with every number traceable to a document.
- Board: the net figure, how long it ran, how it was found and which control failed.
For the board, the "how" matters as much as the "how much." Explaining why the scheme was possible, using a frame such as the fraud triangle of pressure, opportunity and rationalization, helps directors decide which control to change. It does not change the loss figure.
Software helps with the volume. Forensic accounting software and the audit trail it must keep covers what to look for. PreciseBankFlow reads each statement into lines, confirms it foots and exports Excel workpapers with the source lines behind each exception, which gives the schedule above a tested base.
Estimating fraud loss when bank records are missing
When part of the period has no bank statements, the fraud loss for that stretch must be estimated or the records rebuilt. Request copies from the bank first, since banks can usually produce statements and check images for past years. Where nothing can be obtained, estimate from the months you have.
Label any such figure as an estimate in the schedule heading and in the text.
Ways to fill a gap:
- Request the statements from the bank. This is usually possible and is always the first step.
- Rebuild from other records. Check images, the general ledger, vendor payment history and the other party's records can stand in for a missing statement. Reconstructing bank records when statements are missing covers this in detail.
- Extrapolate only as a last resort. If the scheme ran at a steady rate, an estimate based on the average of the known months may be acceptable, but show it in its own column, give the method and expect challenge.
Test the estimate against anything independent you can find. If the subject's own spending, a vendor's sales ledger or a tax return covers the missing months, compare the totals. An estimate that agrees with an outside figure is far easier to defend than one that rests on an average alone.
Never mix rebuilt or estimated months into the confirmed column. A reader who finds one estimated figure presented as confirmed will doubt every other number in the schedule.