On October 19, 2015, Central Board of Direct Taxes (CBDT), issued the final rules1  in relation to multiple year data2  and computation of a range of arm's length prices (applicable to both international and specified domestic transactions). The current notification follows the draft rules3  issued for public comments on May 21, 2015 by the CBDT. This move, prima facie, makes the Indian transfer pricing regulations more aligned with globally accepted best practices on transfer pricing. However, as in most things in India, there are certain unique features of these rules.

Following is the summary of the new rules:

Use of multiple year data

  • The new rules are applicable for transactions undertaken from April 1, 2014 and onwards.
  • The rules provide that data for the current financial year and / or data pertaining to the financial year just preceding the current financial year may be adopted for the purpose of a benchmarking analysis (for comparables).4
  • Use of multiple year data is applicable only if the taxpayer has adopted the resale price method (“RPM”), cost plus method (“CPM") or transactional net margin method (“TNMM”) as the most appropriate method to benchmark the identified transaction (international or domestic) and has considered more than one comparable to benchmark this transaction. 
  • In a case where the data for the current financial year is not available at the time of undertaking the benchmarking analysis, then, the same can be updated and used for determination of the arm’s length price/range during the course of audit proceedings. For example, if while preparing the transfer pricing documentation for 2014-15, the data for current year is not available, then the tax office (during the audit) has the right to update the financial data for 2014-15 and re-compute the range.
  • A  comparable, which is initially part of the transfer pricing documentation (based on the financial information of the year preceding the current year), shall be rejected, if the same is found to be non-comparable, once the data for the current year is available during the course of audit proceedings.

Arm’s length range

  • Application of a range is allowed in the following cases:
  • If the taxpayer has adopted the comparable uncontrolled price (“CUP”) method, RPM, CPM or TNMM as the most appropriate method for the purpose of the benchmarking analysis; and
  • If the comparable dataset has six or more comparable transactions/third parties.
  • Hence, ranges are not applicable if the ‘profit split method’ or ‘other method’5  has been adopted as the most appropriate method.
  • The weighted average of the multiple year data points can be used to calculate the range in cases in which the RPM, CPM or TNMM is the selected method.  As regards the CUP method, the data points for the current financial year only are selected for the computation of a range (as the benefit of using multiple year data is not available under this method).
  • The range is defined as stretching from the 35th percentile to 65th percentile of the data distribution. In case the transfer price does not fall within the computed range, then the median (50th percentile) is to be adopted as the arm’s length price and the difference between the median and the transfer price may be the potential transfer pricing adjustment. 

The 35th percentile and the 65th percentile are computed in the following manner:

  • Step 1 - Place the data points in an ascending order;
  • Step 2 - 35th percentile = The position in the data array corresponding to the total number of data points*(35/100),6 counting up from the lowest observation;
  • Step 3 - 65th percentile = The position in the data array corresponding to total number of data points*(65/100),7 counting up from the lowest observation;
  • Step 4 - When the result from step 2 and/or step 3 is not a whole number, then the comparable price/margin corresponding to the next higher value is to be selected to construct the range;
  • Step 5 - When the result from step 2 and/or step 3 is a whole number, then the comparable price/margin to be used to construct the range is the average of
    (i) the data point corresponding to the computed whole number; and
    (ii) the data point corresponding to the next higher whole number.
  • When the comparable data set has more than one comparable but less than six comparables, then the arm’s length price is arrived at by computing the arithmetic mean of the data points of such comparable data set.8

The new rules also provide illustrations and are explained below:

Illustration - Multiple year data 

Click here to view table.

Illustration - Application of range

Click here to view table.

The range shall be computed as follows:

  • Total number of observations = 7
  • Position of the 35th percentile is 7*(35/100) = 2.45.  In this case, because the position is not a whole number, the next larger value,  the third observation, is selected as the lower bound of the range, i.e., 8.2%.
  • Position of the 65th percentile is 7*(65/100) = 4.55, which is rounded up to the fifth position, and the value of  10.6% is selected as the upper bound of the range.
  • Thus, the arm’s length range is from 8.2% to 10.57%
  • Data place of the 50th percentile (median) is 7*(50/100) = 3.50.  In this case, the value of the fourth observation is selected i.e., 9.0%
  • In situations in which the transfer price does not fall within the computed arm’s length range, the median, i.e. 9%, shall be considered as the arm’s length price.

Comments

  • In relation to the above illustration, the arm's length price (as per the existing regulations) would have been 8.55% (i.e., the arithmetic mean of the data points) as opposed to 9.0% (the median). However, it is important to note that the scale of difference between the median and the arithmetic mean would depend on the dispersion in the data set as the arithmetic mean can be influenced by extreme values.
  • The new rules construct an arm’s length range where the starting and ending point of the range is a data point from the dataset (which is an actual price / margin of a comparable). In some jurisdictions, the arm’s length range is computed based on an inherent formula (say in Microsoft Excel) where the starting and ending point of the range may not be a point from the data set (which is an actual price / margin of a comparable) but a point which lies between the various uncontrolled prices selected based on interpolation.
  • Computation of the 35th percentile and the 65th percentile (arm’s length range) based on the formula in Excel determines the range from 8.28% to 10.41% (as opposed to 8.2% to 10.57%). Thus, for the given example, the arm’s length range, as computed using the Indian methodology under the new rules, provides a wider arm’s length range as compared to practices in some other jurisdictions (computed through Excel based formulas).  Also, the start and end of the Indian range are actual profit margins earned by uncontrolled enterprises, as opposed to the 8.28% and 10.41% values, which are points between the uncontrolled prices.
  • The Indian rules provide a restricted range as compared to a number of jurisdictions where a range extending from the 25th percentile to the 75th percentile of the data distribution is accepted. Hence, for the given example, a different tax authority would have accepted a range between 7.10% to 11.24% (based on the Excel “quartile” formula).  As a result, taxpayers and tax authorities applying the range extending from the 25th percentile to the 75th percentile of the data distribution would have found more transfer prices to be consistent with arm’s-length dealings than the Indian authorities would have, even with identical comparables.

Conclusion

  • Guidance provided in the new rules with respect to the usage of multiple year data and the introduction of the range concept is a welcome change and should help to reduce disputes in the country.
  • Such changes should also influence advance pricing agreement negotiations.
  • CBDT has helped taxpayers to interpret the rules by providing simple examples.
  • The concept of a range provided in the new rules (35th percentile to 65th percentile) is generally narrower than the ranges accepted by other jurisdictions that rely on the interquartile range (extending from the 25th percentile to the 75th percentile). The new rules are, however, still a notch better than the draft rules, wherein it was proposed to have the 40th to 60th percentile as the range.
  • The overall message seems to be consistent that Indian Government wants to reduce transfer pricing litigation and gels well with other recent changes that the Government has proposed on selection and procedural aspects11  of transfer pricing audits in India.
  • A possible next step in relation to transfer pricing documentation in India (for CBDT) could be to provide relaxation on the frequency carrying out search for comparables on databases. The general practise in India is to carry out the entire search for comparables every year. As per OECD's12 final report on Action 13 (transfer pricing documentation and country-by-country reporting), it is recommended that as long as operating conditions do not change, such search for comparables in databases can be done every three years rather than annually. The financial data (for comparables) should be updated every year.13