How to build a transfer pricing local file that holds up.

01.What a local file is actually for

The OECD's three-tier structure, introduced through BEPS Action 13 and set out in Chapter V of the Guidelines, splits documentation into a master file, a local file, and a country-by-country report. The master file describes the group as a whole. The CbC report gives tax authorities a jurisdiction-by-jurisdiction view of where revenue, profit, tax, and people sit. The local file is the document that deals with the specific entity and its specific controlled transactions — and it is the one an examiner reads first when they start asking questions about a particular company.

The United States sits slightly apart from this structure. It adopted country-by-country reporting, but its documentation standard runs through the penalty regulations at §1.6662-6 rather than a formal local file requirement. In practice the content converges: a U.S. study built to the §6662 standard and an OECD local file cover the same ground, and the analysis below applies to both.

Whatever the jurisdiction, a local file has three readers. A tax authority's risk team reads it to decide whether to audit at all. An examiner reads it to decide what to ask. And the group itself should read it, every year, to check that the story still matches the business. Most files are written only for the first reader, and only to the standard of "present and complete."

A file built to hold up is written for the second reader: someone intelligent, unfamiliar with your business, and looking for the gap between what the document says and what the company actually does. The structure that follows has five sections, each asking one question. If the file fails any of them, you have found the place an examiner will start.

Step one: Organizational structure — who really decides?

Chapter V's local file annex asks for the entity's management structure, a local organization chart, and a description of the individuals to whom local management reports, along with the countries where they are based. Most files satisfy this with a chart and a paragraph.

What an examiner does with it is more pointed. The organizational section is where they test whether the characterization of the entity is plausible. A company described as a limited-risk distributor, whose general manager reports locally and sets customer pricing without reference to the principal, has an organizational structure that contradicts its transfer pricing. The chart does not need to say so. The examiner will draw the conclusion.

So describe decision-making, not just reporting lines. For the decisions that matter — pricing, product range, inventory levels, customer credit, hiring senior staff — say who makes them, by role, and where that person sits. Where decisions are made abroad and implemented locally, say that explicitly and describe how.

The test: list the five most consequential business decisions the entity is involved in. For each, can the file tell a stranger who decides and where? If not, the characterization is unsupported.

Step two: Controlled transactions — where contract meets conduct

This section describes each material category of controlled transaction: what is bought, sold, licensed, or provided; the amounts; the counterparties; and the contractual terms. The Chapter V annex also calls for copies of all material intercompany agreements concluded by the local entity.

That last requirement matters far more than its placement suggests. Under the accurate delineation framework in Chapter I, Section D.1, the written contract is the starting point for identifying the transaction — but where the parties' actual conduct differs from the contract, the conduct determines what the transaction is. An agreement that allocates inventory risk to the principal is worth very little if the local entity writes off obsolete stock against its own profit and loss.

Three failures recur. The agreement exists but was never signed. The agreement was signed years ago and no longer describes the arrangement. Or there is no agreement at all, and the file describes the transaction from the tax team's understanding of it. Each one hands the examiner the argument that the arrangement is whatever the evidence of conduct says it is.

"A local file is not a record of what you filed. It is the first draft of your audit defense."

Reconcile the amounts, too. The transaction values in the local file should tie to the intercompany ledger and, through it, to the statutory accounts. A controlled transaction reported at one figure in the local file and another in the financial statements is a small inconsistency that generates a large number of questions.

Step three: Functional analysis — specific, not generic

The functional analysis is the core of the file. It describes the functions the entity performs, the assets it uses, and the risks it assumes, and it is the basis for everything downstream: the characterization, the tested party, the method, and the comparables. As I set out in the benchmarking pillar, it is also where most documentation is weakest.

The weakness is almost always generality. A functional analysis that describes "sales and marketing activities" and "customer support" in terms that would fit any distributor in the industry has not described this company. It has described a category. Examiners read it as a sign that the analysis was drafted from a template rather than from the business.

Risk deserves particular attention. The OECD framework in Chapter I asks not only which party contractually bears a risk, but which party controls it — makes the decisions to take it on and to respond to it — and whether that party has the financial capacity to absorb it. A functional analysis that allocates risk by contract alone, without addressing control, is incomplete against the current standard and easy to challenge.

The test: delete the entity's name from the functional analysis and show it to someone who knows your competitors. If they cannot tell which company it describes, it is not specific enough.

Step four: Method selection and the financial analysis

This section selects the transfer pricing method, identifies the tested party, applies the method, and concludes on the arm's length result. The standards here are the ones covered throughout the benchmarking cluster: the method must be selected by comparison with the alternatives, not assumed; the tested party must be the less complex entity; and the comparables must be screened on criteria set before the results were known.

The failure specific to local files is in the financial data. Where the tested transaction is only part of the entity's business, the analysis usually relies on segmented financials — a profit and loss statement for the relevant activity carved out of the entity's full accounts. The Chapter V annex asks for information showing how the financial data used in the analysis ties to the annual financial statements.

That reconciliation is frequently missing, and its absence is a serious gap. A TNMM result computed on a segment P&L that nobody can reconcile to the audited accounts is a result the examiner has no reason to accept. Cost allocations into the segment deserve the same scrutiny: if the allocation keys are undocumented, the segment margin is an assertion rather than a calculation.

Step five: Contemporaneity — the timing that decides penalties

A local file's value depends heavily on when it was completed. In the United States, penalty protection from the §6662(e) transfer pricing penalties — 20 percent for a substantial valuation misstatement, 40 percent for a gross one — requires that the documentation exist when the return is filed and be provided to the IRS within thirty days of a request. A study completed after the return, however rigorous, does not qualify.

The OECD's position is framed as best practice rather than a penalty condition: the local file should be finalized no later than the due date of the tax return for the year concerned. Many jurisdictions go further, with statutory filing deadlines, submission windows measured in days, or penalties for late or incomplete documentation that apply regardless of whether the pricing was arm's length.

Contemporaneity is not only about the calendar. A file completed on time that describes last year's business is contemporaneous in form and stale in substance. The Guidelines expect the business description, functional analysis, and comparables to be reviewed annually, with the financial data of comparables updated each year even where the underlying search is refreshed less often. Updating the numbers without rereading the narrative is how the controller in the opening scenario ended up with a file describing a warehouse that no longer existed.

02.Eight gaps examiners find first

If your local file has any of the following, fix it before an information request arrives rather than after:

• The business description is out of date. It refers to facilities, products, people, or structures that no longer exist, or omits a restructuring that happened in the period.
• The agreements are missing or unsigned. Material intercompany agreements are cited but not attached, were never executed, or no longer describe the arrangement.
• The transaction amounts do not reconcile. Controlled transaction values in the file do not tie to the intercompany ledger or the statutory accounts.
• The functional analysis is generic. It describes a category of business rather than this entity, and allocates risk by contract without addressing control.
• The characterization contradicts the organization. The file describes a routine entity whose local management makes the decisions a routine entity would not.
• The method selection names no alternatives. The chosen method is stated rather than selected, with no documented reasoning for rejecting others.
• The segmented financials cannot be traced. The tested party's result rests on a segment P&L with no reconciliation to the audited accounts and no documented allocation keys.
• The file contradicts the master file or CbC report. The entity's functions, headcount, or profitability are described differently in the group-level documents than in its own local file.

03.A final word

The controller in the opening scenario had thirty days. She spent them rewriting rather than refiling: correcting the business description, getting the intercompany agreement signed with an accurate description of the arrangement going forward, rebuilding the functional analysis around how decisions were really made, and refreshing the comparables. The file that went to the examiner was better than any version that had existed before.

It was also produced under pressure, at the worst possible time, by someone who had to explain why the prior versions said something different. That is the real cost of the binder problem. The work gets done eventually. It just gets done in the least favourable conditions.

A local file that is read once a year by the person who updates it will, sooner or later, be read closely by someone who wants to disagree with it. Write it for that reader. The discipline is the same every year — describe the business as it actually operates, tie the numbers to the accounts, attach the agreements that govern the transactions, and finish it before the return is filed. Done that way, the file stops being a compliance cost and becomes what it was always meant to be: the first draft of your defense.