Cumulative translation adjustment, CTA-E, and unrealized FX

One exchange-rate movement can produce three different figures in a consolidated set of books, and they are constantly mistaken for each other because they all look like “an FX number”. They are not variants of one thing. Each answers a different question, each lands somewhere else, and only one of them ever touches the income statement. This is the distinction between the cumulative translation adjustment (CTA, and sometimes mis-typed CTE), its elimination sibling CTA-E, and an unrealized FX gain or loss.

What is a cumulative translation adjustment (CTA) in accounting?

A foreign subsidiary keeps its books in its own functional currency, and those books are already correct. Consolidation does not fix them — it presents them in the parent's reporting currency. Under ASC 830 that presentation uses a different rate for each account class: the closing rate for balance-sheet accounts, the period average for the income statement, and historical rates for contributed equity.

Those rates cannot all agree with one another, so the translated balance sheet does not foot on its own. The residual is the cumulative translation adjustment. It is not an error and it is not a plug in the pejorative sense — it is the arithmetic consequence of translating one set of books at several rates on purpose, and every consolidated balance sheet under a foreign subsidiary carries one.

CTA vs unrealized FX gain or loss: translation vs remeasurement

An unrealized FX gain or loss is a different operation on a different object. Remeasurement happens inside one entity's own books, before consolidation is anywhere in the picture. It applies when a monetary balance is denominated in a currency other than the functional currency of the books it sits in: a US-functional company holding a Canadian-dollar receivable, for instance. The amount owed is fixed in Canadian dollars; what it is worth to the US company moves with the rate, and that movement is a real change in the company's own economic position. It belongs in income.

The shortest version: remeasurement asks what a balance is worth; translation asks what a company looks like. Remeasurement changes your position. Translation changes only the yardstick.

A worked example: one rate move, three answers

Take a US parent reporting in US dollars, with a Canadian subsidiary whose functional currency is the Canadian dollar. The rates for the period: 0.78 when share capital was contributed, 0.755 as the period average, and 0.74 at period end. All figures below are generic illustrations.

Remeasurement. The US company holds a CA$1,250 customer receivable booked mid-period at 0.75, so US$937.50 went on the books. At period end the same CA$1,250 is worth US$925.00. The US$12.50 difference is an unrealized FX loss. It posts to the FX gain / (loss) account against accounts receivable, and it reverses into the following period, because the mark states a position as of a date rather than accumulating. When the customer pays, the difference between the booked rate and the settlement rate becomes a realized loss and stops being an estimate.

Translation. The Canadian subsidiary was funded with CA$300,000 of share capital when the rate was 0.78, and earned CA$100,000 during the period. Its net assets at period end are CA$400,000.

• Net assets at the closing rate: CA$400,000 × 0.74 = US$296,000.
• Share capital at its historical rate: CA$300,000 × 0.78 = US$234,000.
• Period earnings at the average rate: CA$100,000 × 0.755 = US$75,500.
• Equity by component therefore totals US$309,500, against net assets of US$296,000.

The gap — US$13,500 — is the cumulative translation adjustment. Nothing was lost and nobody made a mistake. Three rates were applied to one company, exactly as the standard requires, and this is what they leave behind.

What is CTA-E, the elimination residual?

CTA-E belongs to the same family as CTA — it is translation, not remeasurement — with one decisive difference: it is a real posted equity leg. It has to be. The intercompany elimination is a document, and a document has to foot. When the two translated sides of an intercompany pair do not cancel, something has to absorb the difference or the entry cannot post at all.

Each leg is valued from its originating subsidiary's functional amount and translated at that subsidiary's consolidated rate to the common parent. Two sides, two rates, and no guarantee the two rates are the same number. Say the Canadian subsidiary carries a CA$1,250 intercompany receivable and the US subsidiary carries the mirror payable, booked at 0.75 as US$937.50. Neither side has anything to remeasure — each booked in its own functional currency. But at consolidation the Canadian leg translates at 0.74 to US$925.00 while the US leg is already US$937.50. They no longer cancel. The US$12.50 remainder posts to CTA-E, and the elimination entry foots.

CTA-E also absorbs a second, entirely different thing: any asymmetry that was genuinely booked, where one side recorded more than the other. A well-built consolidation keeps the two labelled apart, because a translation residual is expected and a booked asymmetry is a data error. The useful mental model is:

CTA-E = (the translation gap from currency and rate differences) + (any asymmetry that was booked)

Remove the currency difference and the first term dies. The second survives until the source documents are corrected — which is why an out-of-balance pair is always fixed at the source, never by adjusting the elimination.

When CTA-E is exactly zero

Whether a pair can produce a translation residual at all is decided by the currencies involved, not by whether anyone made a mistake. There are three cases, and knowing which one you are in tells you immediately whether a non-zero figure is expected behaviour or a signal to go looking.

1. Different functional currencies. A Canadian subsidiary trading with a US subsidiary. The gap between the period-end consolidated rates and the rates the documents were booked at drives CTA-E. Expect a non-zero figure. Any booking asymmetry sits on top of it and is a separate matter.

2. Both subsidiaries in the parent's own currency. Two US subsidiaries under a US parent. Every edge's rate derives to 1.0, so there is no rate-difference remainder that could exist. On a clean mirror — US$5,000 receivable against US$5,000 payable — CTA-E is exactly zero. If US$150 shows up, that is not translation. One side booked US$5,000 and the other US$4,850, and the fix is in the source documents.

3. Both subsidiaries share a currency that is not the parent's — two Canadian subsidiaries under a US parent — or the pair's legs sit on accounts with different rate types. This is the case that surprises people. Balance-sheet legs translate at the same current spot and still cancel cleanly. Average rates, however, are volume-weighted from each subsidiary's own transaction activity, so two subsidiaries in the same currency can legitimately carry different average rates. On a CA$60,000 intercompany management fee, an average of 0.7550 on one side and 0.7538 on the other produces US$45,300 against US$45,228 — a US$72 CTA-E that is purely the spread between two consolidated rates. No currency mismatch on the documents, no error, and nothing to correct.

Why CTA sits in equity and not income

Because no transaction occurred. Nothing was bought, sold, borrowed, or settled. The subsidiary's operations were identical whether the rate moved or not; only the rate used to present them changed. Running that through net income would report a profit nobody earned and make reported earnings a function of currency markets rather than of the business. So it accumulates in other comprehensive income and stays there for as long as the investment is held.

It does eventually reach income, but only on the way out: when the parent sells or substantially liquidates its investment in the foreign entity, the amount accumulated over the holding period is released and recognized as part of the gain or loss on disposal. That is the one moment the translation actually became economic.

Two habits follow from all of this, and both catch people out. First, a consolidated statement is not the sum of the standalone statements — that difference is the CTA, by design. Second, CTA is specific to the node it was calculated for: a parent's CTA is not its children's CTA plus something, and adding two nodes' adjustments together produces a number that means nothing.

How Cairn computes CTA

In Cairn, the three figures are kept structurally apart rather than netted into one line. Revaluing foreign-currency balances is a step of the month-end close and posts a real auto-reversing journal pair into income. CTA is computed at read time, per consolidating node, in the consolidation snapshot — the subsidiary's own books are never touched — and a translation-adjustment audit decomposes it per account into carried forward, opening retranslation, and translated movement, tying to the cent or telling you plainly that it does not. CTA-E lands on its own equity account so an elimination residual is never read as pure translation, and the intercompany reconciliation reports a booked asymmetry in a separate column from the translation adjustment, so the two are never conflated.

That is the whole point of separating them: three questions, three answers, and an audit trail that says which is which. It is part of the same double-entry general ledger every operational document already posts to, and it is built for the finance teams who have to sign the consolidated numbers off.

Frequently asked questions