What has to go in a local file?
The entity's controlled transactions, the functional analysis, the method selected, the comparables and the financial results. Those five parts are not a checklist to be filled in separately, they are an argument: the functions and risks you describe should be the reason the method makes sense, and the method should be the reason those comparables are the right ones. A file where each section is individually defensible but the sections do not follow from one another is the version that reads badly to an auditor, because the join is where the questions start.
Our margin is below the tested range — what do we put in the file?
The explanation, and you want to be the one giving it. Where the entity's actual margin sits outside the tested range, the file has to explain the difference before the auditor does. That means identifying what caused it — a start-up period, an exceptional cost, a capacity problem, a customer loss — and evidencing that cause from the accounts rather than asserting it. A file that reports a result outside the range and says nothing about it has handed the first question of the audit to the other side, along with the silence that makes it look unanswerable.
Do we need a local file for a management fee we pay our parent?
A charge from a parent is a controlled transaction, so it belongs in the analysis whatever it is called. Management fee is a label rather than a description, and the work is establishing what was actually supplied, by whom, and on what basis the amount was set. That is what the functional analysis has to carry. The fees least likely to survive a review are the ones where nobody can say what service was received, because the file then has a method and comparables sitting on top of a transaction it has not described.
How old can our benchmarking study be?
The file has to be evidence about the year it covers. A range computed several years ago is evidence about those years, so carrying it forward unexamined leaves a gap between what the file claims and what it shows, and that gap widens as the comparable set ages, companies leave it and the market moves. The practical approach is to check each cycle whether the comparables still fit the entity being tested and whether their own results have moved, and to record that check. Refreshing the search is cheaper than explaining why you did not.
Can we just use the group master file instead of a local file?
No, because they answer different questions. The master file describes the group; the local file is where the group's story meets the local entity's numbers, and it is the document a local auditor tests line by line. The group description contains nothing about this entity's controlled transactions, comparables or results, so it cannot be tested against them. The two documents do need to agree with one another, and where a local file has to be written quickly the master file is the right place to start, but it is not a substitute.
What if an audit starts and we have no local file?
You then build the analysis under time pressure and with the auditor's questions setting the agenda, which is the expensive way round. The file's value is that it fixes the description of the entity's functions, the method and the comparables before anyone is arguing about the result. Prepared afterwards, the same analysis looks assembled to fit an answer, and any inconsistency with the group's own description is found by the person examining you. If an audit has opened, the order of work matters: establish what is being asked, then document.
Can you give a plain transfer pricing example?
A Canadian company manufactures at a cost of one hundred and its US subsidiary sells to customers for one hundred and eighty. If the parent invoices the subsidiary at one hundred and ten, most of the margin is taxed in the United States; invoice at one hundred and seventy and most of it is taxed in Canada. Nothing about the business changed — only which treasury collects. That is why the arm's length price, the one unrelated parties would have agreed, is the reference point both authorities use. See our transfer pricing work.
Is "fund transfer pricing" the same thing as transfer pricing?
No — and if you came here to calculate FTP, this is not it. Fund transfer pricing is a bank's internal allocation of funding costs and benefits between its own business units, a treasury and asset-liability management discipline used to measure branch or product profitability. Tax transfer pricing is about prices between legally separate related parties across borders, and about which country taxes the resulting profit. The words overlap; the fields do not. See our transfer pricing work.
