The transfer pricing math needn’t be so ‘mean’ after all

Parikshit Datta and Sanjay Tolia

Arm’s length price should be determined based on multi-year data and by using appropriate systems

The Finance Act, 2001 introduced the Indian transfer pricing (TP) laws as a part of the Income Tax Act, 1961 (Act) largely based on the Organisation for Economic Cooperation and Development (OECD) TP guidelines with a detailed set of rules for their implementation. The law has been in force for seven years now and the companies are in the fourth year of TP assessments in India.

Several issues have come up during the assessment proceedings, which may lead to prolonged litigation in some cases. One such issue, which merits attention, is the selection of the data period for analysing the comparability of an uncontrolled transaction with a controlled transaction, together with the use of arithmetic mean to arrive at the arm’s length price.

The Indian legislation requires a comparability analysis to be based on data pertaining to the year in which the taxpayer undertakes the international transactions with associated enterprises (current year data). However, the tax payer may also use data of the past two years where such past data reveals facts that could influence the determination of transfer prices of the current year.

During TP audits though, the TP officers have generally tested the taxpayer’s result using current year data of comparable companies, ignoring earlier year data. The issue attains further importance in the light of the fact that under the Indian TP legislation, arm’s length price of an international transaction is determined having regard to the arithmetic mean of all the comparable transactions with a limited flexibility of a 5% range around the mean.

The Indian TP laws being in their seventh year of enforcement, it may be time we took cognisance of the issue and tried to align the same with the international practices.

The OECD TP guidelines prescribe use of multiple year data and state that for a complete understanding of the facts and circumstances surrounding the controlled transaction, it generally might be useful to examine data from both the year under examination and prior years.

On a closer look at the TP laws of some of the developed countries, like the US and Australia, one would find a similarity with the above noted OECD guideline.

In conducting comparability analysis in business sectors which are volatile in nature, it may be a better approach to consider the results of comparable transactions/margins over a period of time, rather than consider data for a single or current year. This approach, more importantly, applies to emerging economies like India, which are experiencing significant growth in certain industries, while certain industries are maturing over time and moving towards a more stable profit regime. Even within high-growth industries, profits could be highly volatile over years and not follow a systematic trend. Use of multiple year data in such a case would even out the variances on account of yearly fluctuations and provide a more reasonable measure of the arm’s length outcome.

Coming to the question of determination of the appropriate statistical measure for arriving at the arm’s length price, the OECD guidelines state that when a TP analysis produces multiple results, all of which are relatively equally reliable, use of a range may be particularly appropriate and accordingly, an arm’s length range may be established.

Most of the developed countries with established TP practice consider median as an appropriate statistical measure of central tendency and the inter-quartile range (IQR) for determining arm’s length range.

However the Indian TP legislation is not explicitly clear on the use of multiple year data for the purpose of comparability analysis and also does not adopt median and the IQR to determine arm’s length prices. This creates a paradox of its own when the arm’s length price is determined based on the current year data by adopting arithmetic mean of all the comparable transactions with a limited flexibility of a 5% range around the mean.

For this purpose one needs to analyse ‘arithmetic mean’ and the range envisaged in the Indian legislation of 5% around the mean vis-à-vis the median and the IQR.

Median is the middle most value usually determined by sorting the data set from lowest to highest values.

In a given set of arm’s length prices or margins, an IQR represents the spread between the 75th percentile and the 25th percentile.

As median is the middle most value and IQR uses the middle 50% of data, neither of them are affected by extreme values in the given set of prices/margins.

Compared to this, the arithmetic mean is simply the sum of all the arm’s length prices/margins in the set, divided by the total number of items in the set. Since arithmetic mean uses the values of the entire data points present in the sample, the mean is influenced by the extremes of the data set. Therefore, the use of arithmetic mean to arrive at an arm’s length price and thereafter determining the arm’s length range by allowing a limited flexibility of 5% around the mean may not be the right approach.

Consequently, it may be stated that the use of current (single) year data instead of an average of past few years data of comparable companies (as well as the taxpayer’s) in combination with adopting the arithmetic mean with a limited flexibility of 5% around the mean, may not provide an appropriate indication of the arm’s length range of prices/margins.

Such practice completely ignores the commercial and economic conditions surrounding the transaction, fails to even-out product and business life cycles or specific economic conditions that may impact some companies and distorts the arm’s length price so arrived at.

Within an industry, a comparable company may incur losses or earn super profits in any one year due to a variety of reasons. However, its business may stabilise over a few years and considering data relating to multiple years of its operations may provide a more correct picture of its profitability.

On the contrary, if the Indian TP law would have adopted median and IQR as statistical measures of arm’s length price and range, respectively, the extreme values may not have that impact as they in most cases would have been excluded, unlike in the case of arithmetic mean, even though the calculation of median and IQR would have been based on current (single) year data.

Having regard to the above it may be worthwhile for the Indian legislators to explicitly state the use of multiple year data in conjunction with an arithmetic mean approach to determine arm’s price to avoid any unwarranted litigation or to revisit the current legislation and adopt the concept of median & IQR with multiple year data to align it with the international best practices.

Sanjay Tolia is executive director and Parikshit Datta associate director of PricewaterhouseCoopers.