Do I really need a benchmarking study for intercompany service charges?
If the charge exists, the question is not its size but whether you can show the price was set on arm's length terms. A benchmarking study is the usual way to show it, because it puts an independent range beside your own result. Where the charge is routine and low risk, the study can be short: a clear description of the tested party, a documented search, and a range. What you cannot do is assert that the mark-up is reasonable and leave it there. An assertion is not evidence, and on review it is treated as though nothing had been prepared.
Can I reuse last year's benchmarking study for this year?
Often yes, provided nothing has changed in what the tested party does. The search strategy and the screening decisions can stand while the financial data behind them is refreshed. What forces a new search is a change in the business: a new function taken on, a risk moved, a different set of customers, a restructuring inside the group. Reusing a study through such a change is the weakest position of all, because the document now describes an entity that no longer exists. Record the review each year, even in the years where the conclusion is that nothing needed to change.
Why has the auditor asked for my list of rejected comparables?
Because that is where the range was actually decided. The database search produces a long list of candidates, and the study you file reflects a much shorter one. Everything that happened in between — the quantitative screens, then the manual review of each remaining company — is what moved the range up or down. An auditor who cannot see those rejections cannot test the result, so the natural step is to rebuild the set themselves and compare. A study with a full rejection log invites a conversation about individual companies. A study without one invites a conversation about whether the range means anything.
How many comparables does a benchmarking study need to be defensible?
There is no count that makes a study safe. What matters is that the search was wide enough to start with, the screens applied to it were stated, and each company that survived the screens was looked at and either kept or rejected for a written reason. A small final set built that way is defensible. A larger set assembled by loosening the screens until the range covered the tested party's own result is not, and the loosening usually shows up in the audit trail. Judge the study by whether a reviewer could repeat it, not by the size of the final list.
What if there are no comparable companies in my own country?
That is common for smaller markets, and it does not stop the analysis. The usual route is to widen the geographic screen to a broader region and then say so plainly: which countries were included, why they were treated as an acceptable market for this function, and what the effect of the widening was. The alternative — forcing a range out of two or three local companies of doubtful comparability — is weaker than a well-reasoned regional set. The reasoning has to be in the document. A search that silently spans several countries reads as an unexplained search.
Can I prepare the benchmarking myself from public financial statements?
You can look at published accounts, and for a rough sense of where your margin sits that is worth doing. What that exercise cannot produce is a documented search: there is no record of the population you started from, no stated screening criteria, and no rejection log, so nobody can tell whether the companies you found are the comparable ones or simply the ones you happened to see. A study is a method plus a record of applying it. If the result may have to be defended, the search itself has to be reproducible by someone who was not there.
What are the transfer pricing methods?
Five, in two groups. Three compare transactions: comparable uncontrolled price, resale price, and cost plus. Two compare profits: the transactional net margin method, and profit split. The OECD asks for the most appropriate method on the facts rather than a fixed hierarchy; the United States applies a best-method rule to similar effect. Selection is itself a documented judgment, and a method chosen without recording why is a weak position under audit. See our transfer pricing work.
What is an intercompany agreement, and do we need one?
It is the contract between the related parties — who does what, who bears which risk, what is charged and on what basis. It matters because when there is no agreement, an auditor prices the transaction from the conduct they can observe rather than from the arrangement you intended, and conduct rarely tells the whole story. Signed agreements that match the invoices and the actual functions are the cheapest transfer pricing protection there is. See our transfer pricing work.
