Skip to content

Direct Tax Laws & International Taxation · Transfer Pricing

Comparability Analysis, Adjustments and Range Concept in Transfer Pricing

Updated 5 October 2026 · Fact-checked

Comparability analysis tests whether an uncontrolled transaction is similar enough to the international transaction to price it. You pick comparables, adjust for differences, then use the arithmetic mean (fewer than six comparables) or the 35th to 75th percentile range with the median (six or more) to fix the arm's length price.

Understand Comparability Analysis, Adjustments and Range Concept

Transfer pricing asks one question: would unrelated parties have agreed this price? You cannot answer it without a yardstick. The yardstick is a comparable uncontrolled transaction or a comparable independent company. Comparability analysis is the work of choosing that yardstick and proving it is fair.

Two transactions are comparable when no difference between them materially affects the price or margin, or when reasonable adjustments can remove the effect of the difference. You judge this on the comparability factors: characteristics of the property or services, functions performed (taking into account assets used and risks assumed), contractual terms, and economic and market conditions such as geography, market size, level of competition and timing. A simple way to remember it: what is sold, who does what and bears which risk, what the contract says, and where and when it happens.

No two companies are identical. So you make comparability adjustments for reliably measurable differences, such as working capital (receivable, inventory and payable days), credit terms, volumes, capacity utilisation and, where relevant, country risk. Adjust only when the difference is real, measurable and affects price or margin. If you cannot measure it reliably, drop that comparable rather than guess.

Multiple year data means using data of years other than the year of the transaction. The general rule is to use data of the year of the transaction. The transfer pricing rules allow data of not more than two years prior to the year of the transaction, and only when that earlier data reveals facts that could influence the determination of the transfer price in the transaction you are examining, such as a product life cycle or a start-up phase. Do not use it routinely. This was Rule 10B(4) in the earlier rules; verify the corresponding rule in the Income-tax Rules, 2026.

Once you have comparable prices or margins, you must fix a single ALP. Where the most appropriate method gives fewer than six data points, the ALP is the arithmetic mean. The actual price is accepted if the variation between the ALP and the price at which the transaction was actually undertaken does not exceed the notified tolerance band, which is a percentage of the actual transaction price. Apply the band to the price of the transaction, not to profit or sales. Where there are six or more, the range concept applies: arrange the data in ascending order, find the 35th and 75th percentiles, and accept the price if it lies within that range. If it lies outside, the ALP is the median. Note that India uses the 35th to 75th percentile range, not the 25th to 75th interquartile range used in some OECD practice. Check the exact thresholds and the tolerance percentage in the transfer pricing provisions of the Income-tax Act, 2025, the rules made under it and the latest notification.

Key rules to remember

Comparability factors
Product/service characteristics + Functions (assets, risks) + Contractual terms + Economic and market conditions
Use these four heads to structure every comparability answer.
Arithmetic mean rule
ALP = Σ (comparable prices or margins) ÷ n, when comparables < 6
Accept the actual price if the variation between the ALP (the mean) and the price at which the transaction was actually undertaken does not exceed the tolerance band notified by the Central Government (commonly 1% for wholesale traders and 3% for others; verify the current notification). Variation test: |ALP price − actual price| ÷ actual transaction price. The base is the actual transaction price, never sales or profit. If the mean is a margin, first convert it into an ALP price, then test.
Range concept (percentile rank)
Rank = P% × n, rounded up to the next whole number, in the ascending series; P = 35, 50 (median), 75
Applies when comparables ≥ 6. A percentile is the lowest value such that at least P% of the data is less than or equal to it.
Range test
If 35th percentile ≤ actual price ≤ 75th percentile → price accepted as ALP; else ALP = median
The adjustment is then (median − actual price) × quantity, or the equivalent on margin.
Working capital adjustment
Adjustment to comparable's margin = [(WC of tested party ÷ base) − (WC of comparable ÷ base)] × interest rate; WC = receivables + inventory − payables; base = sales or cost, matching the margin
Add to the comparable's margin if the tested party carries more working capital; deduct if it carries less.
Multiple year data
Current-year data by default; data of not more than two years prior to the year of the transaction only if it reveals facts that could influence the determination of the transfer price
State the reason for using earlier data in your answer. This was Rule 10B(4) in the earlier rules; verify the corresponding rule in the Income-tax Rules, 2026.

How to solve Comparability Analysis, Adjustments and Range Concept questions

Use this order for any question on comparability, adjustments or ALP determination from a set of comparables.

  1. 1Identify the international transaction, the tested party and the most appropriate method (CUP, RPM, CPM, PSM, TNMM) from the facts.
  2. 2List the comparability factors for each comparable: product or service, functions, assets and risks, contract terms, market conditions. Reject comparables with differences that cannot be adjusted.
  3. 3Decide whether current-year data is enough, or whether earlier-year data reveals facts that could influence the price. Give the reason in one line.
  4. 4Compute comparability adjustments for measurable differences, such as working capital. Apply them to the comparable, not the tested party, and show the working.
  5. 5Count the comparables after rejections. Fewer than 6: take the arithmetic mean. Six or more: arrange in ascending order.
  6. 6For the range, compute ranks for the 35th, 50th and 75th percentiles (P% × n, rounded up) and read off the values.
  7. 7Compare the actual price with the range, or with the mean and the tolerance band. For the band, convert the mean into an ALP price if needed, then measure the variation between the ALP and the actual transaction price as a percentage of the actual transaction price, not of sales or profit. If the price is accepted, no adjustment is made. If not, the ALP is the median (six or more comparables) or the mean (fewer than six comparables).
  8. 8State the adjustment in rupees (difference × quantity, or the difference between ALP price and actual price) and the conclusion.

Quickest way: Rank, Read, Compare

When to use it: Use this when the question gives a list of comparable prices or margins and asks whether the transaction is at arm's length.

  1. Write the data in ascending order first. Never work from the given order.
  2. Count n. If n is less than 6, take the mean and go straight to the tolerance band check (variation between the ALP price and the actual transaction price, as a percentage of the actual transaction price, not of sales or profit).
  3. If n is 6 or more, multiply n by 0.35, 0.50 and 0.75. Round each up to the next whole number. Those are your ranks.
  4. Pick the values at those ranks. Check whether the actual price lies between the 35th and 75th values, both ends included.
  5. If it is outside, take the median value as the ALP and multiply the gap by the quantity.

Common mistakes in Comparability Analysis, Adjustments and Range Concept

  • Using the 25th to 75th percentile (interquartile range) as the Indian range.

    OECD material and many textbooks use the interquartile range, and students carry that over.

    Fix: Remember that India's range runs from the 35th to the 75th percentile. Mention the interquartile range only as a contrast.

  • Averaging the two middle values for the median, or interpolating percentiles.

    Statistics classes teach interpolation.

    Fix: Use the rule's definition: the lowest value such that at least P% of the data is less than or equal to it. Compute P% × n and round up to get the rank.

  • Applying the range concept with fewer than six comparables.

    Students remember the range but forget the threshold.

    Fix: Count the comparables first. Fewer than six means arithmetic mean and the tolerance band.

  • Adjusting the tested party instead of the comparable, or applying the working capital adjustment in the wrong direction.

    The sign logic is not understood, so students guess.

    Fix: Ask what the comparable's margin would be if it had the tested party's working capital. If the tested party has more working capital, the comparable's margin goes up.

  • Adjusting for differences that are not reliably measurable, or using earlier-year data as a routine.

    Students think more adjustments and more data always make the analysis better.

    Fix: Adjust only for measurable differences that affect price. Use earlier-year data only when it shows facts that influenced the current year's price, and say why.

  • Taking the whole difference from the mean as the adjustment when the price is within the tolerance band.

    The tolerance band step is forgotten.

    Fix: Always test the actual price against the mean and the notified tolerance band before computing any adjustment.

Worked examples

Example 1

An Indian company sold 10,000 units of a component to its foreign associated enterprise at ₹99 per unit. Using the CUP method, seven comparable uncontrolled prices per unit were found after adjustments: ₹104, ₹96, ₹112, ₹100, ₹98, ₹108 and ₹102. Determine the ALP and the adjustment, if any.

Show the solution
  1. Arrange in ascending order: 96, 98, 100, 102, 104, 108, 112. n = 7, which is at least 6, so the range concept applies.
  2. 35th percentile rank: 0.35 × 7 = 2.45, rounded up to 3. The 3rd value is ₹100.
  3. 75th percentile rank: 0.75 × 7 = 5.25, rounded up to 6. The 6th value is ₹108.
  4. Median rank: 0.50 × 7 = 3.5, rounded up to 4. The 4th value is ₹102.
  5. The range is ₹100 to ₹108. The actual price of ₹99 is below ₹100, so it is outside the range and not accepted.
  6. ALP = median = ₹102. Adjustment = (₹102 − ₹99) × 10,000 = ₹30,000.

Answer: The ALP is ₹102 per unit. The income is increased by ₹30,000. Had the price been, say, ₹105, it would be within ₹100 to ₹108 and accepted with no adjustment.

Example 2

An Indian contract manufacturer (tested party, not a wholesale trader) incurred total operating cost of ₹50,00,000 and charged its foreign associated enterprise ₹54,50,000, an operating profit of 9.0% on cost. Its working capital (receivables + inventory − payables) is 20% of operating cost. TNMM is the most appropriate method, with operating profit to total cost as the profit level indicator and two comparables. Comparable A: margin on cost 10%, working capital 12% of cost. Comparable B: margin on cost 9%, working capital 24% of cost. Use a working capital interest rate of 10% and a tolerance band of 3%. Compute the adjusted margins and the mean, convert the mean into an ALP price, and decide whether any adjustment is needed.

Show the solution
  1. Comparable A: difference = 20% − 12% = 8%. Adjustment = 8% × 10% = 0.8%. The tested party has more working capital, so add. Adjusted margin = 10% + 0.8% = 10.8%.
  2. Comparable B: difference = 20% − 24% = −4%. Adjustment = −4% × 10% = −0.4%. Adjusted margin = 9% − 0.4% = 8.6%.
  3. Only two comparables, fewer than 6, so the range concept does not apply. Use the arithmetic mean: (10.8% + 8.6%) ÷ 2 = 9.7% on cost.
  4. Convert the mean margin into an ALP price: ₹50,00,000 × (1 + 9.7%) = ₹54,85,000. The actual price is ₹54,50,000. The difference is ₹35,000.
  5. Tolerance test: the base is the actual transaction price, not profit or sales. Variation = ₹35,000 ÷ ₹54,50,000 ≈ 0.64%.
  6. The variation of about 0.64% does not exceed the 3% tolerance band, so the actual price of ₹54,50,000 is accepted and no adjustment is made. Had the variation exceeded 3%, the ALP would be ₹54,85,000 and the adjustment ₹35,000.

Answer: The adjusted comparable margins are 10.8% and 8.6%, with a mean of 9.7% on cost. This gives an ALP price of ₹54,85,000 against the actual price of ₹54,50,000. The variation is about 0.64% of the actual transaction price, which is within the 3% tolerance band. The actual price is accepted and no adjustment is made.

Exam tips

  • Write the four comparability factors as a short list in any theory question. It earns easy marks and structures the answer.
  • In numerical questions, always show the sorted series and the rank calculation. Marks are awarded for the method even if the final figure is off.
  • Check the number of comparables before choosing between mean and range. Case scenarios often include one comparable that must be rejected, which can change the count below six.
  • State the direction of every adjustment in words, such as 'tested party has higher working capital, so comparable margin is increased'.
  • For MCQs on this topic, the common traps are the 35th to 75th range (not 25th to 75th), the threshold of six comparables, and the median being the ALP when the price falls outside the range.

Practice questions from Transfer Pricing

Comparability Analysis, Adjustments and Range Concept in other exams

The same ground in other exams, if you are preparing for more than one or want another angle on it.

Comparability Analysis, Adjustments and Range Concept: frequently asked questions

What is the range concept in transfer pricing?

When you have six or more comparable data points, you arrange them in ascending order and find the 35th and 75th percentiles. A price within that range is accepted as arm's length. A price outside it is replaced by the median.

How do I compute the 35th percentile in the Indian method?

Multiply 0.35 by the number of data points and round up to the next whole number. The value at that rank in the ascending series is the 35th percentile. For 7 comparables, 0.35 × 7 = 2.45, so you take the 3rd value.

Is the interquartile range used in India?

No. Some OECD guidance uses the interquartile range of the 25th to 75th percentile. India's range runs from the 35th to the 75th percentile, and the median replaces the price when it falls outside.

When can I use multiple year data?

The default is data of the year of the transaction. You can use data of not more than two years prior to the year of the transaction, and only if it reveals facts that could influence the determination of the transfer price. This was Rule 10B(4) in the earlier rules; verify the corresponding rule in the Income-tax Rules, 2026. Say clearly why the earlier data is relevant.

How does a working capital adjustment work?

You compare the working capital of the tested party and the comparable as a percentage of the same base as the margin, usually sales (or cost if the margin is on cost). Multiply the difference by an interest rate. Add the result to the comparable's margin if the tested party carries more working capital, and deduct it if it carries less.