The transactional profit methods (the comparable profits method or CPM in the US and Taiwan; the transactional net margin method or TNMM almost everywhere else) shifts from a standard in which transfer prices are fixed based on direct observation – e.g., at the same level of a price for the same product when sold between two unrelated parties – to one in which they are inferred based on other information – e.g., the profits earned by a comparable company or a comparable transaction. The theory underlining the CPM/TNMM is that if controlled transactions have the same prices as uncontrolled transactions, then the profits generated by the controlled transactions are, by definition, the same as they would be had the parties to the transaction paid (received) the third party prices. Therefore, all else equal, the profits realized in the controlled transaction should be the same as those realized in uncontrolled transactions. As it is easier to directly observe the profits generated in uncontrolled transactions than to observe the prices that are charged, setting transfer prices at a level that gives controlled entities the same profits as can be observed from uncontrolled entities should result in prices that are similar to the (unobservable) prices that exist in the third party transactions.

However, there are issues with the application of this simple theory in practice. To start with, there are two potential sources of profit data (the profits of the seller, the profits of the buyer), and the answer that is derived from the application of a transactional profits method will depend upon whether the profits of the seller or buyer are used as a starting point. Given this, the determination of whether the seller or buyer will be the “tested party” used as the starting point for the analysis becomes important.

Second, given that the profits of both the tested party and the comparables selected as profit benchmarks are affected by both (i) the price charged form the specific transaction at issues and (ii) other things, transfer prices set using a CPM/TNMM will be affected by factors that would not have had any impact upon prices charged at arm’s length. The reliability of the CPM/TNMM is therefore dependent upon the relative impact of the controlled transactions and “other” things on profits.

If 99% of the tested party’s costs (or revenues) are derived from the prices paid in the controlled transactions and 99% of the comparables costs (or revenues) are derived from comparable transactions, transfer prices are normally more important than “other things”, and the CPM/TNMM is likely to be a reliable approach to inferring prices.

If 1% of the tested party’s costs (or revenues) are derived from the controlled transactions and 99% are derived from “other things”, the CPM/TNMM is unlikely to be a reliable way of inferring prices even if you can find comparables where 99% of their costs (or revenues) are derived from comparable transactions. Simple statistical noise makes any inferences based on profits unreliable – a 1 percentage point change in profits will lead to a 100% change in the transfer prices for the controlled transactions.

If 99% of the tested party’s costs (or revenues) are derived from controlled transactions but only 1% of the comparables costs (or revenues) are derived from controlled transaction, then the reliability of the CPM/TNMM depends very much on the economic assumption that the business activities of the tested party and those of the comparables should be equally profitable.

[There are also lots of issues around the reliability of different PLIs. But that is a topic for another day.]

Somewhat interesting, the guidance provided in both the US transfer pricing regulations and the OECD Transfer Pricing Guidelines approach the question of the reliability of the CPM/TNMM analysis indirectly by talking about the importance of reliable data and the need to use the “simpler” party to the transaction. However, they do not explicitly address the core issue that the ability to make reliable inferences about prices from profits is clearly dependent upon the existence of a close relationship between the two.

Because profits are affected by both the prices charged in the controlled transactions and “other things”, the use of the CPM has led to a blurring of the distinction between the treatment of profits that are related to current business operations and profits that are based on investment activities.

As a simple example, a company may decide to take an accounting write-off of capital equipment used by the tested party because of a drop in demand for the product. At arm’s length, such a decision is unlikely to affect prices – it deals with the accounting treatment of a fixed cost – and is likely to occur in an environment where prices may be falling in reaction to the reduction in demand. In the CPM/TNMM world, however, the write-off leads to an increase in current costs, which in turn leads to an increase in the price that is needed by the tested party to meet the “routine” level of profits observed in the comparable data.

As a more extreme result, the widespread adoption of CPM/TNMM as the primary pricing method has led to an expectation on the part of certain tax authorities that subsidiaries must earn an arm’s length profit regardless of the level of intercompany transactions. Thus, it is now relatively common to see the CPM/TNMM used in situations where intercompany transactions account for only a minor share of the costs or revenues of the tested party, and therefore have only a limited impact upon profits. It is not all that unusual to see tax authorities setting CPM/TNMM target that would, for example, require the seller to pay the buyer for the sale of products, even though such negative prices are rarely if ever observed at arm’s length.

In essence, the widespread use of transactional profits methods has led to a shift in the focus of the arm’s length standard from the requirement that prices be set at the same level as they would in a transaction between two controlled parties to a standard that requires that the profits of “routine” transactions or “routine” entities be set at the same level as those observed in comparable transactions/entities, even if this results in profits that are clearly non-arm’s length (e.g., a seller paying a buyer to take goods that are produced in the normal course of business). This in turn has led to situations in which transfer pricing rules have been used to, in effect, guarantee and/or limit returns for investment activities, even when the profits of these investments are not dependent upon intercompany transactions.

As a caveat, I would note that the above discussion applies to the behavior and expectation of tax authorities as well as some MNEs and transfer pricing professionals, and not necessarily to the specific regulations in place in various countries. At least in the US, I personally do not believe that legal standards have changed from their focus on arm’s length prices per se.

Blog Posts in this Series:

Evolution of the Arm’s Length Standard: Introduction

Evolution of the Arm’s Length Standard: Part 2 – Entering a Market Through Sales v Investment

Evolution of the Arm’s Length Standard: Part 3 – The Introduction of Profit-Based Methods

Evolution of the Arm’s Length Standard: Part 4 – Reasonable Alternatives

Evolution of the Arm’s Length Standard: Part 5 – An Increasing Focus on the Location of Decision-Makers


Please subscribe to my blog if you are interested in receiving my new posts.