Transfer Pricing – Use of Range and Multiple Year Data for Determining Arm’s Length Price

Two of the areas where Indian transfer pricing regulations (‘TPR’) glaringly deviated from international best practices were use of inter-quartile range and multiple year data for benchmarking. Till less than a year ago, the TPR suggested the use of arithmetic mean of multiple comparable prices with a deviation of a small percentage (a mere 5%, further brought down to 1% in case of wholesalers and 3% for other assessees) for calculation of the arm’s length price (‘ALP’). It also did not permit the use of multiple year data, unless such data was shown to have a bearing on the determination of the transfer prices.

These provisions were challenged time and again in practice and before the appellate authorities, but in vain. While the learning curve of the Indian legislature and the tax authorities in transfer pricing has been quite steep, the need for use of range of arm’s length prices and multiple year data was acknowledged only in the finance minister’s budget speech in 2014, followed by necessary amendments to the law, applicable to transactions undertaken on or after 1st April 2014, i.e., from A.Y. 2015- 16 onwards. The enabling rules for the same, however, were notified only on 19th October 2015.

Need for the amendment:

Multiple year data –
Although broadly comparable companies are identified based on the functions, assets and risks (‘FAR ’) analysis, their profits will be a result of the returns normally commensurate with such FAR , coupled with any factors affecting the returns of the industry as a whole, as well as the returns of the individual company. The company specific factors, say merger, demerger or any other corporate action, which are normally restricted to a single year, will result in the rejection of an otherwise comparable company. Alternatively, these factors will distort the profits of that particular comparable, impacting the overall ALP calculation.

Assuming no material variation in the FAR of the comparable companies, the use of multiple year data helps to eliminate or “iron out” the impact, if any, of the extraordinary factors occurring in any of the years, yielding the normal profit margins of the industry and thereby, producing more reliable results.

Range concept –
While the transfer pricing methods try to objectively compute the ALP, it will seldom be the case that the actual price of the transaction will exactly match a single ALP. Deriving arithmetic mean in case of multiple comparable prices, with a small range of deviation, implies that the transaction price must nearly match with the mean, often leading to inequitable results. On the contrary, as per the international practice of adopting the inter-quartile range, the transaction price has to be compared to a range of prices between the start of the second quarter and the end of the third quarter of the comparable prices, when arranged in an ascending order.

For instance, say eight comparable prices – P1 to P8 are obtained as a part of the benchmarking exercise. As per the arithmetic mean concept, the simple average of P1 to P8 will be considered to be the ALP, with an allowable deviation of 1/3%, as the case maybe. However, as per the concept of inter-quartile range, these eight prices will be divided into 4 quarters and the entire range of second and third quarter (i.e., P3 to P6) will be considered to be arm’s length. Evidently, a range of prices is a better measure of the ALP than the arithmetic mean.

Draft Rules:
The draft scheme for applying the range concept and use of multiple year data, inviting comments, was released on 21st May 2015. Its highlights were as under –

Multiple year data –
Applicable only if, and mandatory if, the most appropriate method (‘MAM’) used for computation of ALP is Resale Price Method (‘RPM’) or Cost Plus Method (‘CPM’) or Transactional Net Margin Method (‘TNMM’).

Data for three years, including the year of transaction or current year should be considered.

Data for two out of the three years shall be used in case the data for the current year is not available at the time of filing of return of income, or a comparable does not clear the quantitative filter in any one year, or data is available only for two years on account of commencement or closing down of operations.

If the data for current year becomes subsequently available, the same can be used at the time of the assessment by the assessee as well as the department.

Applies irrespective of whether range concept or arithmetic mean is used.

Range concept –

Applicable only if the MAM used for computation of ALP is RPM or CPM or TNMM.

At least nine comparable companies, based on FAR analysis, should be available.

Weighted average margins of last three years (or two years in certain circumstances), shall be calculated using the denominators of the Profit Level Indicator (‘PLI’) as the weights.

Arranging the above margins in an ascending order, the values between 40th and 60th percentile of such margins shall be considered as the arm’s length range.

If the transaction price is within such range, then no adjustment shall be made. However, if the transaction price is outside the range, then the entire difference between the transaction price and the median or central value of the range shall be adjusted in the income.
In cases where the range concept does not apply, i.e., in case of benchmarking as per Comparable Uncontrolled Price (‘CUP’) Method or Profit Split Method (‘PSM’) or any other method as per Rule 10AB, or where number of comparable companies is less than nine, the earlier computation as per arithmetic mean and the tolerance range shall continue to apply.

Final Rules:
The final rules on the subject have been notified vide Notification No. 83/2015 dated 19th October 2015. The implications of these rules are as under –

Multiple year data –
A second proviso has been added to sub-rule (4) of Rule 10B to provide that the first proviso, dealing with the use of earlier years’ data where it has a bearing on the determination of the transfer prices, shall not apply in case of transactions entered into on or after 1st April 2014.

Also, sub-rule (5) has been inserted to provide that where, in respect of transactions entered into on or after 1st April 2014, the MAM selected is either RPM or CPM or TNMM, then, the comparability of an uncontrolled transaction with the controlled transaction shall be analysed based on the data pertaining to the current year (i.e. the year in which the transaction was entered into, say F.Y. 2014-15) or the immediately preceding financial year (F.Y. 2013-14) if the data for the current year is not available at the time of furnishing the return of income for that year.
Further, the proviso to sub-rule (5) states that if the data relating to the current year (F.Y. 2014-15) becomes subsequently available at the time of assessment of the said year, then, such data shall be used for computation of ALP, even though it was not available at the time of furnishing the return of income. This proviso intends to settle the persisting issue of inappropriateness of the use of that data at the time of assessment, which was not available to the assessee at the time of filing of return of income.

Rule 10CA, inserted by the above notification, deals with the application of range concept as well as multiple year data, in detail along with illustrations. The provisos to sub-rule (2) of Rule 10CA lay down the following mechanism for use of multiple year data –

i)    Where the comparable uncontrolled transaction has been identified using current year (F.Y. 2014-
15) data as per Rule 10B(5), and the comparable entity (and not the assessee) has entered into same or similar uncontrolled transaction in either or both of the immediately preceding financial years (F.Y. 2012-13 and F.Y. 2013-14), then,

•    the price of the uncontrolled transactions for the preceding financial years (F.Y. 2012-13 and F.Y. 2013-14) shall be computed using the same method as is applied for the current year (F.Y. 2014-15), and

•    weighted average price shall be computed by assigning weights to the sales/costs/assets employed or the respective denominator, used in computing the margins as per the MAM.

ii)    Due to non-availability of current year data (F.Y. 2014-15) at the time of filing of return of income, if the comparable uncontrolled transaction has been identified using the data for the immediately preceding financial year (F.Y. 2013-14) as per Rule 10B(5), and the comparable entity (and not the assessee) has entered into same or similar uncontrolled transaction in the immediately preceding financial year of that year (F.Y. 2012-13), then,

•    the price of the uncontrolled transactions for that preceding financial year (F.Y. 2012-13) shall be computed using the same method as is applied for the financial year immediately preceding the current year (F.Y. 2013-14), and

•    weighted average price shall be computed in the same manner as above.

iii)    Further, where data for current year was not available at the time of filing the return of income but was subsequently available at the time of assessment, and it is found that the uncontrolled transaction of the current year (F.Y. 2014-15) is not same or similar or comparable to the controlled transaction, then, that entity shall be excluded from the set of comparables, even if it had carried out comparable uncontrolled transaction in any of the preceding two financial years (F.Y. 2012-13 and F.Y. 2013-14).

In other words, an entity selected as comparable on the basis of comparable uncontrolled transaction entered into in the year preceding the current year (F.Y. 2013-14), shall be outright rejected if it is later found out that it does not have comparable uncontrolled transaction during the current year (F.Y. 2014-15).

The above calculation of weighted average prices of multiple year data will apply in all cases where RPM, CPM or TNMM have been selected as the MAM and comparable uncontrolled transactions are available in the current year as well as any or both of the immediately preceding two financial years, irrespective of whether the range concept or the arithmetic mean is applicable.

Range concept –

As per sub-rule (4) of Rule 10CA, the concept of range shall apply in case of transactions where the MAM selected is not PSM or any other method and where the dataset of prices of comparable uncontrolled transactions consists of at least six entries. The entries in the dataset will be the weighted average prices of the comparable uncontrolled transactions where RPM, CPM or TNMM was selected as the MAM and comparable uncontrolled transactions were also entered into during the preceding financial years. In other cases, i.e., where CUP is selected as the MAM or where no comparable uncontrolled transactions are available in the preceding financial years, the prices calculated using the MAM for the current year will form part of the dataset.

To apply the range concept, the dataset has to be first arranged in an ascending order and the prices starting from the thirty-fifth percentile and ending with the sixty-fifth percentile shall be considered to be the arm’s length range. If the transaction price is within the above arm’s length range, it shall be considered to be at arm’s length. However, if the transaction price is outside the range, then, the median or central value or fiftieth percentile of the dataset will be considered to be the ALP and the difference between the transaction price and such ALP shall be the amount of adjustment.

As a corollary to sub-rule (4), the range concept shall not apply where the dataset has less than six entries or where PSM or any other method is selected as the MAM. In such cases, sub-rule (7) states that the existing computation of arithmetical mean of the values in the dataset and tolerance range of 1/3%, as the case may be, will apply.

The chart on the next page, summarises the provisions relating to range concept and use of multiple year data – (In the chart, Year 3 refers to the current year)

At the end of Rule 10CA, three illustrations have been provided to explain the calculation of weighted average price, selection and rejection of comparable where current year data is not available and calculation of percentile. These are self-explanatory and hence, are not covered in this article.

Issues:

Some of the issues that arise from the rules are as set out below the chart .

Chart: Use of Range concept and Multiple year data

Rule 10B(4), prior to the amendment, provided that only data pertaining to the year, in which the transaction has been entered into, should be used for the purposes of benchmarking, unless earlier years’ data has a bearing on the determination of transfer prices. As per the second proviso to Rule 10B(4) now inserted, the provision relating to use of earlier years’ data shall not apply to transactions entered into after 31st March 2014. Further, sub-rule (5) provides for use of data of current year or immediately preceding financial year in case of transactions entered into after 31st March 2014, where RPM, CPM or TNMM is selected as the MAM. Thus, it appears that earlier years’ data cannot be used for transactions entered into after 31st March 2014, where CUP, PSM or Rule 10AB has been selected as the MAM, even though earlier years’ data is shown to have a bearing on the transfer prices.

ii)    Chaos during assessments:

The use of data relating to immediately preceding financial year at the time of filing return of income and use of current year data at the time of assessments will invariably lead to changes in comparables. Consequently, if the number of comparables reduces below six, or a different MAM is adopted during the course of the assessment, the ALP computation may show wild variations, especially due to parallel usage of range concept and arithmetic mean. This will only increase the uncertainty surrounding transfer pricing.

iii)    Acceptance of consistent loss making companies: It is nearly a settled principle that companies that are consistently loss making cannot be accepted as comparable companies since it indicates improper functioning or inefficiencies or discrepancies, etc.

Similarly, several tribunals have held that high profit making companies should also be rejected. With the use of multiple year data, the year on year aberrations are meant to even out. Would it then imply that loss making or high profit making companies can be accepted? It is worth noting that an entity with losses in two out of three years has been accepted as a comparable illustration 1. Logically, consistently loss making or high profit making companies will still need to be excluded, as these entities will have operational differences that cause the substantial deviations, which in turn will translate into differences in FAR.

Conclusion:

The attempt to align the Indian TPR with international practices by introducing provisions for use of the long debated range concept and multiple year data is a welcome move. The final rules are even slightly more liberal as compared to the draft scheme released a few months ago. However, it appears that the complicated drafting of the rules and possibility of re-doing the entire benchmarking process during assessment may end up doing more harm than good.