Transactional profit indicators for the TNMM are not viable

By Dr. Ednaldo Silva, Founder, RoyaltyStat, Washington DC area

It is not viable to determine “transactional” profit indicators for the so-called transactional net margin method (TNMM) in transfer pricing analysis.

Profit indicators that are disaggregated per product or per related party activity are not available for the tested party or for the potential comparables. 

In fact, the OECD should delete the word “transactional” from the TNMM in the transfer pricing guidelines.

Oligopoly pricing

First, consider how products are priced.

Multinational enterprises tend toward market concentration by horizontal and vertical integration (mergers & acquisitions), such that in time a few enterprises grow to control the selected market.

In business history, we observe two concurrent trends: first, individual MNEs become larger; second, the given market becomes concentrated among a few dominant oligopolies selling differentiated products with close substitutes.

Given pervasive concentration, market price is not a datum. Market concentration was visible already in Europe by 1838, when French mathematician Augustin Cournot (1801-1877) developed his enterprise-based profit margin equation.

 The more elegant presentation of Cournot’s equation, written by Amoroso (1930), equation 15, p. 222, is as follows:

(1) (p_i − c_i) / p_i = µ_i (q_i / Q)

for i = 1 to N enterprises.

The variable p_i is the market price, c_i is the average cost and expenses, µ_i > 1 is the profit margin, and q_i is the quantity supplied by the i-th enterprise in the selected market. The coefficient µ is the reciprocal of the elasticity of demand.

In Amoroso’s profit margin equation (1), the variable 0 < (q_i / Q) ≤ 1 is the market share of the i-th enterprise. The number of enterprises determines the market structure in which competition and monopoly are the limits:

Q = ∑ q_i = q₁ + q₂ + q₃ + … is the sum of the quantities supplied by single product enterprises.

In standard microeconomic theory, the product market is composed of many small enterprises, and none can influence the market price; thus, in received theory N approaches infinity such that the market share of any single enterprise is minuscule (irrelevant for price determination).

This large N economic fiction is a tool of miseducation because competition is a Macondo reality.

Competition (large N) and monopoly (N = 1) are limits to the Cournot and Amoroso market conception; economic reality is dominated by oligopolies.

The Cournot-Amoroso markup price equation can be obtained from the profit margin relation (1), as follows:

(2) p_i = λ_i c_i

where (abstracting from subscripts) the markup parameter λ = (1 − µ η)⁻¹ and η = (q / Q) is the market share.

Amoroso’s price equation (2) cannot be estimated in practice because average cost and expenses are not accounted per product.

Product costs

Unless we assume a unicorn single product enterprise, which is an unrealistic assumption, far from the reality of products exchanged cross-border between controlled enterprises, only certain costs and expenses can be attributed to single products.

Vide, the individual accounts constituting cost of goods sold, operating expenses, and depreciation and amortization are not reported disaggregated per product or per related and unrelated party activity.

The tax and accounting concepts of costs of goods sold (COGS) is different from the economics ideal concept of variable costs; only certain accounts in COGS reported on Form 1125-A can be attributed to specific products.

For example, beginning and ending inventories may be attributed to specific products (goods are a misnomer because they can be harmful), but trial balances do not report per product account details. Inventories can be valued according to various methods, which are not always disclosed on attached tax filing schedules or reported accounting footnotes.

Purchases in COGS can be attributed to specific products, but trial balances do not report per product account details.

Labor costs can be attributed to specific products, but trial balances do not report per product account details.

Other costs in COGS are filed according to amorphous schedules, and often we find large sums with the duplicate name “other costs.” In our experience, “other costs” in COGS cannot be reported per product because they can include unattributable deductions such as the amortization of property, plant and equipment used in production.

In SEC-filings, COGS are reported as a single aggregate figure for the potential comparable enterprises, which may or may not be accompanied by accounting footnotes. Thus, COGS are aggregate accounts not disclosed per product.

In transfer pricing analysis, if the “tested party” can produce trial balance COGS accounts per product, those details cannot be obtained from comparables, making this onerous exercise futile.

Advocates’ pleas that transfer pricing analysis must be “transactional” (disaggregated per product or disaggregated per related party activities) are ignorant of tax and SEC-filing reporting standards.

The plea that COGS can be separated between related and unrelated activities is farcical because some accounts in COGS, such as depreciation and amortization, are not separable. 

Operating expenses

Operating expenses are not a concept of Form 1120, US Corporation Income Tax Return, or its relatives.

The accounting concept of operating expenses is vague and may include (or exclude) depreciation and amortization of property, plant and equipment employed below gross profits activities, such as in those activities related to business administration, research & development, advertising, and marketing.

To our knowledge, none of the separate accounts constituting operating expenses (tax deductions), except for advertising, can be attributed to specific products. None of those accounts, again except perhaps for advertising, is reported in separate, disaggregated form.

Research and development expenses may not be disclosed as a separate account. In Form 1120, research and development is not enumerated as a separate deductible account, and in our experience, this important account (important because it creates self-developed “trade” intangibles) is reported in “other costs” in COGS or in “other deductions” on line 26 of Form 1120.

The argument that total deductions in tax filing (line 27 of Form 1120), or “book” accounting operating expenses in SEC-filings, which are reported as a single aggregate account, can be computed per product, or per “transaction,” or can be separated between related and unrelated activities is naïve or ignorant.

Depreciation and amortization

Depreciation and amortization also cannot be attributed to an individual activity. In tax filings, depreciation and amortization are reported in a complex Form 4562.

In practice, depreciation attributed to production activities are reported as an aggregated figure in “other costs” in COGS. Depreciation of administrative, sales, and marketing activities are reported as an aggregated figure in “other deductions.”

In SEC filings, depreciation is reported ad hoc, and we must read the accounting footnotes to collect depreciation reported into its own account.  Moreover, depreciation and amortization are major accounts that make certain European tax administrations’ insistence on obtaining comparables from privately held enterprise financials databases unacceptable. Among accounting vices, privately held enterprises are not obligated to disclose annual reports, including in accounting footnotes. As a result, we don’t know if profits are reported before or after depreciation, or if they include extraordinary items.

The Cournot-Amoroso price equation – aggregated

The Cournot-Amoroso price equation can be estimated by aggregating both sides using the quantity supplied as weights. As analysts, we know the aggregate variables net sales S = ∑ p_i q_i and total cost C = ∑ c_i q_i.

The constant profit markup remains unaffected by the aggregation of activities (aggregation of individual transactions such as the invoiced p_i q_i).

In practice, we can estimate the profit markup equation in aggregate form per enterprise:

(3) S_t = λ C_t +U_t for t = 1 to T years of enterprise-level data.

where U denotes random uncertainty.

Thus, enterprise-level profits can be measured only in the aggregate because the required data are reported combining the multiple products of the potential comparables.

The claim to separate the related and unrelated activities of the tested party is an unfulfilled dream because important costs and expenses cannot be attributed to individual products or to those segmented economic activities. The same holds when measuring comparable enterprise-level profits for transfer pricing purposes.

In short, enterprise-level profit indicators are by nature aggregate measures and the notion they can be measured per transaction is nonsense (business or geographic segment operating profits are based on attributed operating expenses and may include related party transactions; thus, they are unreliable for transfer pricing purposes).