The Same Transaction, Recorded Twice
One shipment, two companies, two systems. Two records that were both correct when they were made, and a difference that only shows up when somebody adds them together.
A group moves goods from one of its companies to another. Two records get created — a despatch in the sending entity, a receipt in the receiving one — and both are made in good faith by people doing their jobs properly.
They still disagree. Not because anybody is careless, but because the two records were made on different dates, in different currencies, in different systems, by people with different information. Each is defensible on its own. The difference between them is nobody's error, which is precisely why nobody owns it.
Then somebody adds them up.
Said first rather than last: we do not consolidate, and we do not eliminate
No statutory consolidation, no intercompany elimination, no group reporting pack. This post is entirely about what happens before any of that — whether the two underlying records describe the same event. That is where the difference is created, and eliminating a balance whose two halves disagree just moves the disagreement somewhere less visible.
Four ways one event becomes two different facts
| The divergence | Why both sides are right |
|---|---|
| Date | Despatched on the 29th, received on the 3rd. Two periods, one movement. Neither entity is wrong about when it saw the goods |
| Currency | Recorded in each entity's own denomination. If either converted at storage, the original figure is gone and only one of the two can be reconciled to source |
| Rate | Two conversions, done at two moments, possibly from two sources. The gap is arithmetic, not judgement |
| Quantity | What left and what arrived, which on a real journey are sometimes different numbers — and that difference is a fact worth keeping rather than smoothing |
The reason intercompany balances are hard is not that anybody is sloppy. It is that two correct records of one event are still two records.
What makes it worse than an ordinary reconciliation
An external reconciliation has an arbiter. The bank statement is the bank statement; the supplier's ledger is the supplier's ledger, and if you disagree you telephone them. There is a third document, and somebody outside the argument owns it.
Intercompany has no arbiter. Both sides are you. So the difference gets resolved by whichever entity has the more insistent finance lead, or by a journal at the centre that makes the totals agree without either underlying record changing — which is worse, because now three things are recorded and the two originals still disagree.
The fix is one mechanism, and it is unglamorous
A movement between two of your own companies should be one object with two confirmations, not two objects that ought to match.
That single change removes most of the divergence: the despatch and the receipt are two events on one record, the in-transit period is an owned position rather than a gap, and a quantity difference is attached to the movement it happened on rather than appearing as an unexplained variance in one entity's count.
Transfers that require confirmation at the receiving end
One movement, two acknowledgements. The stock is owned and visible for the whole period between them, and a shortage belongs to the leg it occurred on.
Multiple entities and locations on one basis
Each with its own stock, approvals and numbering, and one place to ask a question across all of them — instead of a local package here, spreadsheets there, and something inherited from an acquisition.
Amounts kept in the currency they happened in
With the rate actually applied on the record. This is what makes each side reconcilable to its own source rather than to a converted figure whose origin has been discarded.
No report that silently spans currencies
A single total names its currency. A group figure that blended several is the failure this rule exists to prevent, and it is the one most likely to survive undetected because it always looks plausible.
Statutory consolidation and elimination
Not ours. We make the underlying records agree, which is the part that is usually wrong before anybody gets as far as eliminating anything. The consolidation is your reporting team's and your auditor's.
What an intercompany price should be
Not ours at any price, and not a software question. We record the transaction you entered. What it should have been is a matter for advisers with the relevant expertise and the relevant liability.
The test, at the next month end
- Take one intercompany balance. Can both entities produce the list of transactions behind their side of it?
- For one transaction on that list: are the two records the same object, or two objects that agree?
- What was in transit at the period end, and did both entities agree it was in transit?
- When the balance last did not agree, what fixed it — a correction to a record, or a journal at the centre?
- If the answer is a journal: is the original difference still there, underneath it?
The last two questions are the diagnostic. A group that fixes intercompany differences with central journals is not reconciling, it is hiding a reconciliation — and the cost is that the same difference recurs every period, is re-fixed every period, and nobody ever finds out what caused it.
What is built here, what is not, and what we would decline is on the Singapore market page. The wider version — why a group's operational picture arrives monthly while its financial one is timely — is the finance function is here, the operations are not.
