Balance sheet reconciliation

What reconciled actually means

A balance sheet balance is reconciled when you can produce, for the reporting date, the items that make it up, and when the sum of those items equals the balance. That is the whole test. Not that the account looks reasonable, not that it agrees with last month plus the movement, and not that the trial balance balances. A list of items, adding to the balance, each of which you can support.

This definition is stricter than common practice, and deliberately so. A large proportion of what is filed as a reconciliation is really a proof of movement: opening balance, plus what was posted, equals closing balance. That arithmetic is guaranteed by the ledger. It confirms nothing about whether the opening balance was ever right, and it will carry a ten-year-old error forward indefinitely without ever failing.

The distinction matters most for the accounts nobody looks at. Receivables get attention because customers complain. Cash gets attention because the bank publishes a statement. The accounts that accumulate errors are the quiet ones: other debtors, other creditors, accrued expenses, deposits paid, sundry balances. These are almost always supported by a roll-forward, and the roll-forward is why the error has survived.

The rest of this guide goes through the balance sheet by account class and states, for each, what the supporting evidence is and what a reconciliation for that class has to prove. The general ledger mechanics that produce these balances are assumed; what follows is about substantiating them.

The three kinds of reconciliation, and the one that is a guess

There are only three ways to substantiate a balance sheet account, and knowing which one you are performing tells you how much comfort it gives.

The trap is that a roll-forward can be dressed as a substantiation. A prepayment schedule listing four items with monthly releases looks like a detailed listing. If the four items were carried over from a spreadsheet that was itself carried over, and nobody has traced any of them back to an invoice in years, it is a roll-forward with line breaks. The test is whether each line can be traced to a source document, not whether there are lines.

A practical way to grade your own reconciliations: for each balance sheet account, write down what the evidence would be if an auditor asked for it tomorrow. Where the honest answer is a schedule maintained by finance, with no external or systemic source behind it, that account is unsubstantiated regardless of how neat the schedule is.

• Agreement to an independent third party. The balance is compared with a statement produced by someone outside the business: a bank, a lender, a tax authority, a supplier. This is the strongest form because the counterparty has no incentive to agree with your error.
• Substantiation from a detailed listing. The balance is compared with a sub-listing generated from transaction detail: the aged receivables report, the fixed asset register, the stock valuation, the open purchase order accrual. Strong, provided the listing is generated independently of the ledger balance rather than derived from it.
• Roll-forward from a prior balance. Opening balance, plus movements, equals closing balance. This proves arithmetic and nothing else. It is a valid presentation of a movement, and it is not substantiation.

Cash and bank

Cash is the easiest class to reconcile properly and the least excusable to get wrong, because the evidence is external and arrives on its own. The reconciliation compares the ledger balance to the bank statement balance and explains every difference by an identified item that will clear within a normal cycle.

What makes a cash reconciliation weak is not usually the bank account itself. It is the surrounding cash-equivalent accounts that get treated as an afterthought. Petty cash needs a physical count and a signed record, not a balance. Tills need a daily reconciliation between takings and banked amounts, with the variance recorded rather than absorbed. Card and payment gateway settlement accounts need agreement to the gateway statement, because the difference between what customers paid and what was settled consists of fees, refunds and chargebacks, all of which are real costs arriving without an invoice.

Overdrafts and revolving facilities belong in this discussion too, though they sit on the other side of the balance sheet. They reconcile the same way, to the lender statement, and the accrued interest on them is an accrual that should be calculated from the facility terms rather than estimated.

• Evidence: bank statement, gateway settlement report, counted cash certificate, till z-readings.
• Must prove: every difference is an identified item, with a date and a reason it has not cleared.
• Warning sign: reconciling items older than the clearing cycle, or a residual difference described as timing.
• Warning sign: matches made on amount alone, which will pair unrelated items of the same value.

Trade receivables

The receivables control account should equal the total of the aged receivables listing, invoice by invoice. That is the reconciliation. Anything that appears in the control account but not in the listing is a journal that bypassed the subledger, and it is the single most common cause of a receivables difference.

But agreement to the subledger is only half the work, because the subledger can agree perfectly while the balance is not recoverable. The second question is valuation: is each of these amounts going to be collected? That is where the ageing does its real work. An invoice more than a year old with no payment activity and no dispute correspondence is not a receivable in any meaningful sense, and continuing to carry it at full value overstates both assets and profit.

Credit balances within receivables deserve separate attention. A customer account in credit means either an overpayment, a payment on account not yet matched, or a credit note issued without an offsetting invoice. Netting these against the debit balances of other customers, which happens automatically when only the control account total is examined, understates both receivables and the liability to refund. Most frameworks require the two to be presented separately where they are not legally offsettable.

On the provision itself: expected credit loss models under IFRS require a forward-looking estimate rather than a wait-for-default approach, and the level of sophistication expected varies with the size and nature of the business. What is universal is that the provision should be derived from the actual ageing and actual collection history, and that the derivation should be written down. The judgement rests with the preparer; a system can supply the ageing and the payment history it is based on.

• Evidence: aged receivables listing by customer and invoice, cash receipts after the reporting date, correspondence on disputed items.
• Must prove: the listing agrees to the control account, and each aged item is either collectable or provided against.
• Must prove: unallocated receipts have been matched or identified, not left sitting as a bulk credit.
• Warning sign: a provision calculated as a fixed percentage that has not changed in years while the ageing profile has.

Inventory

Inventory reconciliation has two independent legs, and passing one says nothing about the other. The first is quantity: does the system think you hold what you physically hold? The second is value: is the cost attached to those quantities correct, and is it still below net realisable value?

Quantity is proved by counting. A full count at the reporting date is the traditional answer; cycle counting, where every item is counted on a rolling schedule with high-value or high-movement lines counted more often, is usually better because it finds errors closer to when they happen. Either way, what matters is that count differences are investigated rather than posted. A count adjustment is an answer to nothing; it records that stock is missing without saying why, and a business that adjusts without investigating will keep losing the same stock.

Value is proved by the valuation report agreeing to the ledger, and by the costing basis being applied consistently. IAS 2 measures inventories at the lower of cost and net realisable value, with cost assigned on a first-in first-out or weighted average basis, applied consistently for inventories of a similar nature and use. LIFO is not permitted under IFRS although it remains available under US GAAP. The reconciliation therefore has to demonstrate not only that the numbers agree, but that the basis used to produce them is the stated policy.

Work in progress is the hardest case and the one most often supported by a roll-forward. If your work in progress balance is calculated as last month plus costs added less costs transferred out, with no independent measure of what is actually in progress, that balance is an assertion. It needs a listing of open jobs with their costs to date, agreed to the ledger.

• Evidence: stock valuation report by item, count sheets with variances, purchase invoices supporting recent costs, evidence on slow-moving and obsolete lines.
• Must prove: valuation total agrees to the ledger, quantities agree to counts, and the costing basis matches stated policy.
• Must prove: net realisable value has been considered for slow-moving, damaged or superseded stock, with a written basis for any write-down.
• Warning sign: negative quantities in the valuation, which mean issues were posted before receipts and the cost is fictional.
• Warning sign: goods in transit and consignment stock treated by assumption rather than by contract terms.

Prepayments, deposits and other debtors

These are the accounts where the rolled-forward guess lives. They are individually small, they rarely attract attention, and they are supported by schedules that finance maintains itself with no external source. Over several years they accumulate items that nobody can explain and nobody will write off.

A prepayment reconciliation should list, for each item, the supplier, the original invoice, the total amount, the period it covers, the amount released to date and the balance remaining. Every line should trace to a document. The release calculation should be arithmetic, not judgement: an annual insurance premium releases in twelve equal parts unless there is a reason it should not. When the remaining balance of an item reaches zero it should leave the schedule rather than persisting as a zero line for another three years.

Deposits paid are a different animal and are routinely wrong. Rent deposits, utility deposits, customs bonds and supplier deposits are all recoverable amounts held by someone else, and they should be confirmed against a document from that party rather than carried on faith. A deposit for a property vacated four years ago is not an asset; it is an expense that was never recognised.

• Evidence: original invoices for each prepaid item, contracts stating the period covered, deposit confirmations or lease documents.
• Must prove: each line traces to a document, and the release calculation is arithmetic against the period covered.
• Warning sign: a line item described only by a supplier name, with no invoice reference and no covered period.
• Warning sign: an other debtors balance that changes by small amounts every month and never by large ones, which suggests posting rather than resolution.

Fixed assets

The fixed asset register is the reconciliation. Cost, accumulated depreciation and net book value per the register must equal the corresponding ledger accounts, asset class by asset class rather than in total. Agreement in total can hide two offsetting errors in different classes, and the classes carry different depreciation rates, so the error compounds.

The movements in the year need their own proof: additions traced to purchase invoices and to the capitalisation policy, disposals traced to a sale document with the profit or loss on disposal calculated, transfers between classes explained, and the depreciation charge agreeing to what the register calculated rather than to what was posted. That last one catches a surprising number of errors, because a manually posted depreciation journal and a register-calculated charge diverge quietly the first time somebody rounds.

IAS 16 requires that significant parts of an asset with different useful lives be depreciated separately, and that useful life, residual value and depreciation method be reviewed at least annually. A reconciliation that never questions whether the rates in the register still reflect how the assets are used is complete and unhelpful. Assets fully depreciated but still in use are the usual signal that the review is not happening.

Capital work in progress needs the same treatment as inventory work in progress. It should be a listing of projects with costs to date, each traceable to invoices, with a stated expected completion. A project sitting in capital work in progress for three years is either not going to complete, in which case it should be assessed for impairment, or it completed and nobody transferred it, in which case depreciation has been understated throughout.

• Evidence: the fixed asset register by class, purchase invoices for additions, disposal documents, the depreciation calculation.
• Must prove: register agrees to ledger by class for cost, accumulated depreciation and net book value.
• Must prove: every addition met the capitalisation policy, and every disposal removed both cost and accumulated depreciation.
• Warning sign: a large population of fully depreciated assets still in use, which suggests useful lives are not being reviewed.
• Warning sign: assets in the register that no longer physically exist, which is the reason a periodic physical verification is worth doing.

Trade payables and accrued expenses

Payables reconcile to the aged payables listing the same way receivables do, and they carry the same risk of manual journals bypassing the subledger. The difference is in the direction of the risk. Receivables tend to be overstated because uncollectable items are carried too long. Payables tend to be understated, because the way a payable goes missing is that the invoice was never entered at all, and nothing in the ledger will ever tell you about a document you do not have.

Completeness is therefore the central question for payables, and it is proved by looking outside the ledger. Supplier statements are the strongest evidence available and are underused: a statement from the supplier lists what they think you owe, and reconciling to it finds invoices you never received. The second test is the search for unrecorded liabilities, which means reviewing payments made after the reporting date and asking, for each, whether the cost belonged in the period just closed.

Accrued expenses are the mirror image and the harder half, because they have no subledger. Each accrual needs a basis stated in writing: the contract rate and the days accrued, the open purchase order value less invoices received, a usage report, a fee scale. Goods received not invoiced should be systemic, derived from receipts with no matched invoice, rather than estimated. An accrual whose basis is last month plus a bit is not a reconciliation of anything.

• Evidence: aged payables listing, supplier statements, post-period-end payment listing, contracts and purchase orders supporting accruals.
• Must prove: listing agrees to the control account, and material supplier balances agree to supplier statements.
• Must prove: each accrual has a stated basis and a stated release, and stale accruals have been challenged.
• Warning sign: debit balances in payables, which are usually unmatched payments, duplicate payments or missing credit notes.
• Warning sign: an accrual carried at the same figure for more than a few periods, which usually means the underlying invoice arrived and was posted elsewhere.

VAT and tax control accounts

The VAT control account is one of the few balance sheet accounts with an external, dated and legally binding counterparty: the return you filed. The reconciliation compares the movement on the account for the period with the return, and the closing balance with what is owed to or recoverable from the authority. Differences must be explained by identified items, not by a rounding tolerance.

A second, independent test is worth running every period and rarely is. Take the output tax in the ledger and divide it by the standard-rated sales in the ledger. The ratio should approximate the standard rate, allowing for zero-rated, exempt and out-of-scope activity. Do the same for input tax against purchases. When these ratios drift, the cause is nearly always a tax code applied to the wrong transaction type, and that error accumulates silently between filings because the control account balance still looks like a plausible amount of money.

In Saudi Arabia the standard VAT rate is 15 per cent, returns are filed with ZATCA, and e-invoicing requirements mean the invoice data itself is transmitted in a defined structure. That raises the reconciliation stakes: the return, the ledger and the transmitted invoice data all describe the same transactions, and any two of them disagreeing is a finding. The same principle applies wherever transaction-level reporting is required, which is an increasing number of jurisdictions.

Corporate tax, withholding tax and any local levies follow the same logic: a control account per obligation, reconciled to the filed return and to the payment, with the deferred tax position supported by a schedule of temporary differences rather than by a roll-forward. Deferred tax in particular is an account where the balance is often the balance simply because it was the balance.

• Evidence: filed returns, payment or refund confirmations, the tax transaction listing supporting the return.
• Must prove: the account movement agrees to the return, and the closing balance agrees to the amount owed or claimed.
• Must prove: the effective tax rate implied by the ledger is consistent with the mix of rated, zero-rated and exempt activity.
• Warning sign: a balance that never fully clears after payment, which indicates prior period differences absorbed rather than corrected.

Payroll liabilities

Payroll produces several liabilities that clear on different cycles, and treating them as one net balance is why payroll accounts drift. Net pay owed to employees, statutory deductions withheld and owed to an authority, employer contributions, and accrued leave are four separate obligations with four separate counterparties and four different clearing patterns.

The first three are short-cycle and should clear to nil within days of each payment run. If they do not, either the payroll journal is not agreeing to the payroll report, or the payment posted to a different account than the accrual. Both are easy to find and easy to leave, which is why payroll clearing accounts so often carry a small permanent balance that nobody has looked at since the system was implemented.

Accrued leave is different. It is an estimate of the cost of leave that employees have earned and not taken, and it requires a leave balance per employee and a cost per day of leave, both as at the reporting date. Where end-of-service or gratuity obligations apply, as they do across the Gulf, the same principle holds with a longer horizon and more assumptions: the calculation should follow the statutory entitlement formula, run against actual service dates, rather than being estimated as a percentage of payroll. A system can hold the service dates and the leave balances; the assumptions in the valuation remain the responsibility of the preparer, and material obligations of this kind are commonly the subject of specific actuarial or specialist input.

• Evidence: the payroll report per run, statutory filing confirmations, payment records, leave balance report per employee, service dates.
• Must prove: each short-cycle liability cleared to nil after the corresponding payment.
• Must prove: the leave and end-of-service accrual is calculated from actual balances and service dates, not from a percentage.
• Warning sign: a permanent small balance in a payroll clearing account, which is an unfound difference rather than a rounding.

Equity, reserves and retained earnings

Equity accounts move rarely, and that is exactly why they are reconciled badly. Share capital should agree to the constitutional documents and to the share register, not merely to what it was last year. Share premium, statutory reserves where local law requires them, and any revaluation or translation reserves each need a stated basis and a record of the transactions that created them.

Retained earnings is the account most frequently misused, because it is the one place where an unexplained difference can be posted and never questioned. The reconciliation is straightforward: opening retained earnings, plus profit or loss for the period as reported, less distributions declared, equals closing retained earnings. Anything else appearing in that account is a prior period adjustment, and a prior period adjustment is a disclosable event with a reason, not a plug.

This is the point where a roll-forward is legitimately the reconciliation, provided the movements within it are each identified and supported. The difference between a legitimate roll-forward here and an illegitimate one elsewhere is that equity movements are discrete, documented events with board or shareholder approval behind them. If a movement in retained earnings cannot be traced to a resolution, a declared dividend or the reported result, it should not be there.

• Evidence: share register and constitutional documents, board and shareholder resolutions, dividend declarations, prior year signed financial statements.
• Must prove: opening balances agree to the last signed financial statements, not merely to the ledger.
• Must prove: every movement corresponds to an identified, approved event.
• Warning sign: any posting to retained earnings during the year that is not the closing entry or an approved distribution.

Running this as a discipline

A reconciliation programme is worth more than any individual reconciliation, because it is what stops the quiet accounts from going unexamined for years. The programme is a list of every balance sheet account, with an owner, a frequency, the required evidence, and a reviewer who is not the preparer. Accounts that are material or volatile are reconciled monthly. Accounts that are small and stable can be quarterly, provided somebody decided that rather than defaulting to it.

Two rules make the difference between a programme that works and one that produces files. The first is that a reconciliation is not complete while it contains an unexplained difference, however small; a tolerance for unexplained amounts becomes a permanent balance almost immediately. The second is that ageing is reported, so that an item carried on a reconciliation for six periods is visible as such. Most reconciliation failures are not wrong figures. They are correct figures for items that should have been resolved and were instead carried.

Skyline Nexus can produce the subsidiary listings these reconciliations depend on, the aged receivables and payables, the fixed asset register, the stock valuation and the tax transaction detail, tied to the ledger balances they support. That is a statement about what a system can hold and produce. It is not a statement that your balance sheet is right. No software makes books correct, and the judgement about whether a balance is recoverable, complete or fairly stated rests with the preparer and is tested by the auditor.

If you take one habit from this guide, take the grading exercise: go through your balance sheet, account by account, and write down what evidence you would produce if it were asked for tomorrow. The accounts where the honest answer is a schedule finance maintains itself, with no external or systemic source behind it, are your exposure. They are usually small, they are usually old, and they are where the surprises are.

• One owner per account, named, and a reviewer who did not prepare it.
• Stated frequency and stated evidence per account, decided rather than inherited.
• No unexplained differences carried, and reconciling items reported by age.
• The prior period reconciliation retained, so that a carried item is visibly carried.
• A route to write off resolved-as-irrecoverable items, with approval, so old items can actually leave.