Plain English
A single year's results can be misleading — a comparable might have had an unusually good or bad year for reasons that have nothing to do with normal profitability. Using three years of data (commonly a weighted or simple average) smooths that out and gives a more representative picture of what a company earns through a typical business cycle. It is standard practice in benchmarking, though how exactly you average it, and over what period, still involves judgment.
Technical definition
The use of data covering multiple fiscal years, typically three, for both the tested party and the comparable set, under OECD Transfer Pricing Guidelines Chapter III paragraphs 3.75-3.79, to improve the understanding of long-term business results and reduce the distorting effect of cyclical or exceptional single-year events on the comparability analysis.
Why it matters
Single-year comparisons can produce wildly different, and less defensible, ranges depending on which year is picked. Multi-year data reduces volatility in the range and is generally expected by tax authorities as a matter of course rather than an optional refinement.
How it works in practice
- 01Decide the period to use — commonly three years, matched as closely as possible to the tested party's fiscal year.
- 02Source financial data for comparables over the same period.
- 03Decide on simple averaging, weighted averaging, or year-by-year comparison of the tested party against a range computed for each year.
- 04Exclude years affected by one-off, clearly identifiable events if justified and disclosed.
- 05Apply the same treatment consistently to tested party and comparables.
Worked example
Averaging through a downturn
A distributor's single-year 2023 operating margin was 1.1%, depressed by a regional recession. Using a three-year weighted average (2021-2023) of 3.4%, in line with comparables' three-year weighted average range of 2.8% to 4.9%, the tested party is shown to be within range on a normalised basis, whereas a single-year 2023 comparison alone might have suggested an adjustment was needed despite no change in the underlying pricing policy.
Common mistakes
- Using multi-year data for comparables but single-year data for the tested party, or vice versa.
- Cherry-picking which years to include to move the range favourably.
- Ignoring a comparable's structural change (merger, divestment) that makes averaging across the break misleading.
Audit red flags
- Averaging period inconsistent between the tested party and comparables.
- Selective exclusion of years without disclosed justification.
- A benchmarking study that only ever uses single-year data with no explanation.
Documentation & data
Documents to hold
- Stated averaging methodology and period.
- Year-by-year data for both tested party and comparables, not just the average.
- Justification for any excluded year.
Data you need
- Three to five years of financial statements for tested party and comparables.
- Disclosure of any restructuring events affecting comparability in those years.
Who owns this internally: The advisor or economist preparing the benchmarking study.
Jurisdiction notes
- OECD member states
- Three-year averaging is the de facto standard, though single-year testing combined with multi-year context is also accepted.
- United States
- Section 482 regulations permit but do not mandate multiple-year data and require a business reason if single-year data is used exclusively.
Notes by role
Advisors & consultants
Present both the single-year and multi-year view in working papers, even if only one is used in the final report — it pre-empts an obvious audit question.
Students & job seekers
Be ready to explain why averaging matters using a cyclical industry example, such as commodities or construction.
Frequently asked
- Is three years mandatory?
- No fixed number is mandated by the OECD Guidelines, but three years is the practical norm across most jurisdictions and databases.
- Should the tested party also be tested using multi-year data?
- Generally yes, for consistency, although some jurisdictions test the current year's actual result against a multi-year comparables range.
Sources & status
- Primary source
OECD Transfer Pricing Guidelines, Chapter III, paragraphs 3.75-3.79
OECD, 2022
- Our interpretation
Typical multi-year averaging conventions
This glossary, 2026
Reference material only, not advice on a specific fact pattern. Reviewed 2026-06-30.
Careers
How this shows up in the job
Multi-year averaging mechanics are a common Excel test item in TP recruitment case studies.
Careers in transfer pricing