Loading content...
Loading...
Loading...
Median — The Median is the middle value in a dataset when values are arranged in order.
The Median is the middle value in a dataset when values are arranged in order. In transfer pricing, the median is the 50th percentile of the arm's length range—the point at which half of comparable results fall above and half fall below. When a tested party's results fall outside the interquartile range, tax authorities typically adjust to the median rather than to the edge of the range.
The median is preferred over the arithmetic mean because it is less affected by outliers and extreme values in the comparable set.
The OECD Transfer Pricing Guidelines (2022) address the median in Chapter III. The Guidelines note that where conditions of the controlled transaction do not allow a conclusion that the price or margin is at arm's length, it may be appropriate to use measures of central tendency such as the median, which minimizes the risk of error from unknown comparability defects.
US Treasury Regulations §1.482-1(e)(3) state that when the tested party's result is outside the arm's length range, the IRS may adjust the controlled taxpayer's result to a point within the range—typically the median.
When the Median Matters:
| Situation | Role of Median |
|---|---|
| Tested party within IQR | Median is informational only—no adjustment needed |
| Tested party outside IQR | Median is typical adjustment point |
| Setting intercompany prices | Some taxpayers target median for safety |
| APA negotiations | Median often used as target pricing point |
Calculating the Median:
For odd number of data points: Middle value For even number of data points: Average of two middle values
Why Median Over Mean? The median is resistant to outliers. If one comparable has an unusually high or low PLI, the mean shifts toward that outlier, but the median remains stable. This makes median a more reliable central tendency measure for transfer pricing.
Comparable Set: 9 companies with Operating Margins (sorted):
| Rank | Company | Operating Margin |
|---|---|---|
| 1 | A | 1.5% |
| 2 | B | 2.2% |
| 3 | C | 2.8% |
| 4 | D | 3.3% |
| 5 | E | 4.0% ← Median |
| 6 | F | 4.5% |
| 7 | G | 5.2% |
| 8 | H | 5.8% |
| 9 | I | 8.2% |
Calculations:
Tested Party Result: 2.0% (outside IQR)
Adjustment: Tax authority adjusts from 2.0% to 4.0% (median)
| Measure | Calculation | Use in TP |
|---|---|---|
| Median | Middle value | Preferred adjustment point |
| Mean | Sum ÷ count | Less common (outlier-sensitive) |
| Mode | Most frequent value | Rarely used in TP |
The median represents the central tendency—the most likely arm's length result given imperfect comparability. Adjusting to Q1 or Q3 (the edges) would still leave uncertainty. The OECD recommends the median because it "minimises the risk of error from unknown comparability defects."
Yes, many taxpayers target the median as a conservative approach. Pricing at the median provides maximum buffer against fluctuations that might push results outside the IQR. However, any point within the IQR is technically arm's length—there's no requirement to hit the median.
Being at the median is the safest position—it's the most defensible point in the range. If you're at the median, tax authorities have little basis for adjustment. Auditors may still review your methodology, comparable selection, and documentation, but the pricing outcome is robust.
When there's an even number of data points, the median is the average of the two middle values. For example, with 10 comparables, the median is the average of values at positions 5 and 6.
Most do, but some vary. India adjusts to the arithmetic mean of the IQR range rather than the median. Some jurisdictions allow adjustment to any point within the range. Know your jurisdiction's specific rules.
Yes. The median (Q2) is the 50th percentile of the full dataset. It falls within the IQR (Q1 to Q3) by definition—the IQR contains the middle 50% of data, and the median is the exact middle.
With very few comparables (less than 5), statistical measures become less reliable. The "median" of 3 comparables is just the middle value, which may not be representative. Consider whether your comparable set is sufficient, or document limitations in your analysis.