INSIGHT: Transfer Pricing Analysis of Vertically Integrated Companies

March 27, 2020, 7:00 AM

Vertical integration is an arrangement in which the supply chain of a company is consolidated under the same company. In general, each member of the supply chain produces a different product or service, and the products combine to satisfy a common demand. Vertical integration secures the supplies needed by the firm to produce its product and the market needed to sell the product.

Vertical integration can occur in different ways. A company tends toward “forward” vertical integration when it controls distribution centers and retailers where its products are sold. An example is an oil and gas company that produces oil and gas and distributes through its retailers. “Backward” vertical integration occurs when a company controls subsidiaries that produce some of the inputs used in the production of its products. For example, an automobile company may own a tire company and a metal company. Control of these subsidiaries is intended to create a stable supply of inputs and ensure a consistent quality in their final product. A company can also adopt both forward and backward integration in which case it can control most of its supply chain.

By definition, vertically integrated companies perform multiple functions, each of which may be related to a different part of the supply chain. It is also common for multinational enterprises (MNEs) to place each function under a separate legal entity. For example, in a forward vertical integration, manufacturing and distribution functions may be performed separately by two different entities of the same company. When different functions are performed by separate entities of a company, intercompany transactions take place between those entities as the product moves along the supply chain.

In certain industries, vertically integrated MNEs can have several different functions under separate entities which results in multiple back-to-back intercompany transactions for MNEs. For instance, a manufacturing entity of a chemical company can produce an intermediary chemical, which can be used by another manufacturing affiliate to produce the final product which can then be sold to distribution entities for sale to third parties. When there are such back-to-back related party transactions, transfer pricing analysis of those transactions under the comparable profits method (CPM) or Transactional Net Margin Method (TNMM) may require a selection and analysis of multiple tested parties in contrary to the single tested party approach provided by the Treasury Regulations under tax code Section 482 (U.S. transfer pricing regulations) and the OECD Guidelines. The following example illustrates the issue.


A vertically integrated MNE has 4 entities A, B, C, and D which are located in countries Q, X, Y, and Z, respectively, and perform the following functions:
Entity A – Manufactures an intermediary product PA.
Entity B – Uses the intermediary product produced by Entity A to manufacture product PB.
Entity C – Uses product B to further process it to produce the final product PC.
Entity D – Owns the intellectual property (IP) of the MNE and purchases the final product from Entity C for distribution to third parties.

Figure 1: Example of Vertically Integrated Intercompany Transactions

Further, the entities have the following financials with respect to their intercompany transactions:

(*NCP and OM stand for net cost plus markup and operating margin, respectively.)

The question is how one should test Entity C with respect to Entity C’s intercompany transactions. When the entity does not own any IP and performs only routine functions, the typical approach would be to apply the TNMM / CPM to test whether a company earns arm’s length returns with respect to its intercompany transactions. As described in the US Transfer Pricing Regulations:

“… in most cases the tested party will be the least complex of the controlled taxpayers and will not own valuable intangible property or unique assets that distinguish it from potential uncontrolled comparables

In this example, Entity C has two intercompany transactions i.e. purchase of product PB from Entity B and sale of product PC to the distributor. Since Entity C is a routine manufacturer that does not own any IP, one may consider testing Entity C based on the TNMM. With respect to Entity C’s sales of product PC to Entity D, the one might select Entity C as the tested party since Entity D is the IP owner. In that case, looking at Entity C’s financials in Table 1, it might be concluded that Entity C earns an arm’s length return if a 10% NCP achieved by Entity C is within the interquartile range of NCP returns established by comparable manufacturers.

Further, regarding Entity C’s purchase of products from Entity B, one might select Entity B as the tested party. Entity B also earns 10% NCP and, given that comparable manufacturers earn similar NCP returns, the one might conclude also that Entity C’s purchase of product PB from Entity B is also priced consistent with the arm’s length standard. So, based on this analysis, can one say that intercompany transactions involving Entity C meet the arm’s length standard? The answer may not be “it depends.” The reason is that when there are back-to-back intercompany transactions, as we observe commonly for vertically integrated companies, tested party financials are also typically based on intercompany transactions and if those intercompany transactions may not meet the arm’s length standard, testing the results based on comparables using the CPM/TNMM may not necessarily be reliable indicators.

In the example, although the NCPs of both Entity B and Entity C are within the arm’s length ranges based on comparables, those NCPs are calculated based on cost bases of Entity B and Entity C which are also based on intercompany transactions. In particular, Entity B’s COGS is based on the intercompany price of product PA purchased from Entity A. When selling to Entity B, Entity A only earns 2% NCP; and if 2% is not consistent with the arm’s length standard, the intercompany transactions involving product PA at that price may not be appropriately measured. In such a case, since Entity B’s cost base depends on the price of product PA, the use of the CPM/TNMM for purposes of testing Entity B’s intercompany transactions may not be reliable. Similarly, because Entity C’s cost base is dependent on the price of product PB; using the CPM/TNMM to test the Entity C’s transactions may also not be reliable.

Therefore, when testing an intercompany transaction of an entity that has back-to-back intercompany transactions, one should consider not only the tested entity but also all other intercompany transactions that affect the tested transaction. In other words, transfer pricing analysis may require the selection of multiple tested parties as part of the TNMM. In the next sections, specific cases are considered for illustration purposes.

Case 1 – Routine Intercompany Transactions:

Suppose the intercompany transaction chain only includes routine entities i.e. no entity in the chain is considered to own non-routine intangibles. Suppose there are N entities in the intercompany transaction chain where the final entity in the chain is a distributor and all other entities are different manufacturing entities in a vertically integrated supply chain. Figure 2 depicts the transaction chain.

Figure 2: Routine Intercompany Transactions

Let Tn for every n=1,2,…,N denote the transaction where entity n sells to entity n+1. Suppose also that for any entity n=1,2,…,N in the transaction chain, the cost base of entity n has two components: cost of purchase of product from the previous entity (Sn-1) and the additional cost of manufacturing (Cn). Entity n also applies a markup (Mn) on its total cost base when selling the product to the next entity at a price of Sn = (Sn-1+Cn)(1+Mn).

One can then show the price of transaction Tn, i.e. when entity n is selling to entity n+1, as follows:

Equation 1

As seen in Equation 1 above, the price of transaction Tn (Sn) not only depends on the cost base of entity n and markup applied by entity n but also the cost bases of all entities before entity n as well as the markups applied by all entities before entity n. As a result, in order to be able to test transaction Tn, one may need to test all transactions before transaction Tn. If the testing of all the transactions before transaction Tn is consistent with the arm’s length standard, then the TNMM applied to transaction Tn may also be considered reliable.

Given that all entities in the transaction chain are routine entities, it is also possible to benchmark a transaction considering the transactions following the tested transaction. In other words, Equation 1 is derived by applying the arm’s length standard starting with the first transaction in the chain. However, one can also start with the last transaction in the chain. Specifically, since the last entity, in this example, is also a routine distributor, the price of the last intercompany transaction can be tested by looking at the results for the last entity, and the testing of the preceding transactions may also proceed recursively in this way.

Based on this approach in this example, one may consider the following construction in Equation 2 for the price of transaction Tn as follows:

Equation 2


OM is the arm’s length distribution operating margin for entity N,

SN is the third-party sales price charged by entity N, and

Ci , Mi , and N are defined as before.

Equation 2 shows that transaction Tn may also be tested by testing the transactions that occur following Tn in the supply chain. Specifically, if the transactions following Tn are tested under the CPM/TNMM based on the benchmarked markups for manufacturers and operating margin benchmarks for the distributor, then the use of Equation 2 may also be reliable for testing transaction Tn. In this case, the reliability of this approach to testing transaction Tn may depend on the availability and reliability of information regarding the transactions following entity n.

Both Equation 1 and Equation 2 are based on the same principle that all transactions between the tested transaction and third party are considered which ensures that the cost base or revenue related to the tested transaction are consistent with the arm’s length standard. In these examples, both approaches may require the selection of multiple tested parties.

The only difference between the two approaches is the third-party reference point used in the benchmarking. Equation 1 is derived starting with the first entity in the chain that has purchases only from third parties. Equation 2, on the other hand, is derived starting with the last entity in the chain that sells to only third parties.

Case 2 – Routine Intercompany Transactions and One IP Owner:

It is very common that vertically integrated companies have IP related to their technology and / or tradename / trademark. To account for such a case, consider the same example but with one entity that may be considered to own non-routine IP. Suppose entity n is the IP owner in the transaction chain and other entities are still routine entities. Figure 3 below illustrates this case.

Figure 3: Routine Intercompany Transactions and One IP Owner

Although entity n cannot be tested directly because it is the entity that owns the IP, the approach described above in Case 1 may also be applied here i.e. transactions preceding entity n can be tested based on Equation 1 and transactions following entity n can be tested based on Equation 2.

Case 3 – Routine Intercompany Transactions and Two IP Owners:

Some companies may also have more than one IP owning entities in their supply chains. Figure 4 illustrates this case.

Figure 4: Routine Intercompany Transactions and Two IP Owners

In this example, entities n1 and n2 are the IP owners in the supply chain. The transactions preceding entity n1 and following entity n2 may in this instance be tested based on the same logic as described by Equation 1 and Equation 2, respectively. However, transactions between entities n1 and n2 may need to be benchmarked with reference to additional comparables because benchmarking entities n1 and n2 directly may not be reliable.


The US Transfer Pricing Regulations and OECD Guidelines describe the CPM / TNMM as a one-sided method based on the selection of a single tested party. When MNEs, e.g. vertically integrated companies, have back-to-back to intercompany transactions, traditional application of the CPM / TNMM based on a single tested party may not always be reliable because certain costs or revenues along the supply chain are the results of intercompany transactions themselves. As such, one may consider selecting multiple entities along the MNE’s supply chain for purposes of testing transfer pricing.

This column does not necessarily reflect the opinion of The Bureau of National Affairs, Inc. or its owners.

Author Information

Kerem Toklu is a transfer pricing manager with Deloitte Tax LLP. The author thanks Randy Price, Keith Reams, and Ken Reddix for their insightful comments.

About Deloitte
This article contains general information only and Deloitte is not, by means of this article, rendering accounting, business, financial, investment, legal, tax, or other professional advice or services. This article is not a substitute for such professional advice or services, nor should it be used as a basis for any decision or action that may affect your business. Before making any decision or taking any action that may affect your business, you should consult a qualified professional advisor. Deloitte shall not be responsible for any loss sustained by any person who relies on this article.

Copyright © 2020 Deloitte Development LLC. All rights reserved.

To read more articles log in. To learn more about a subscription click here.