Danish Tax Authority Rejects Application of Return on Assets

National tax authorities have recently prevailed in litigations using questionable applications of certain statements in the OECD Transfer Pricing Guidelines. I earlier noted in an April 2020 piece in Tax Management International Journal how the Nigeria tax authorities abused the Resale Price Method to increase the gross margin of a local affiliate distributing chemicals, as well as how the Panamanian tax authorities similarly abused the Resale Price Method to increase the gross margin of a wholesale distributor of petroleum in a February 2021 piece on the EdgarStat Blog. In both litigations, the functions of the distribution affiliate were less than the functions of the allegedly comparable third-party distributors, which implies that the comparison of gross margins with no adjustment for differences in functions overstated the arm’s length gross margin.

Tax Tribunal, Case No. SKM2018.173.LSR was an April 2018 decision in Denmark involving the determination of an arm's length markup for a foreign contract manufacturing affiliate. The intercompany policy was to afford the foreign contract manufacturing affiliate with a 10% markup over operating costs, while the Danish Tax Authority (SKAT) argued that this markup should be less than 4%. The value of the tangible assets for the contract manufacturing affiliate was 110% of operating costs, so the intercompany policy afforded this affiliate with a Return on Assets near 9.1%. The position of SKAT would be equivalent to a Return on Assets below 4%, which was below the risk-free rate.

The court record redacted several aspects of this litigation including what the multinational produced and the jurisdiction of the contract manufacturer affiliate. In our discussion of the economics of this case we shall assume that a German contract manufacturer assembled high-quality speakers that were designed by the Danish parent and sold to German automobile multinationals such as Audi, BMW, and Mercedes to be included in the car’s music system.

The court record noted that this multinational obtained a unilateral Advance Pricing Agreement (APA) with the foreign tax authority. The German tax rate was 30%, while the Danish tax rate was only 25% during the period under review. This multinational likely wanted to guard against the German tax authority asserting a markup in excess of 10%, but was not worried that SKAT would argue for a much lower markup. As the court record noted, SKAT did argue for a lower markup, which would create double taxation for the multinational.

A Financial Model of the Transfer Pricing Issue

Table 1 presents various potential markups in an illustrative example that assumes the following:

• Operating costs total $100 million; and
• Operating assets total $110 million.

Table 1: Illustration of Alternative Markups for Contract Manufacturing Affiliate

The taxpayer’s intercompany pricing policy afforded the contract manufacturing affiliate with a 10% markup over operating costs. Since operating assets represent 110% of operating costs, this policy provided the affiliate with a 9.09% return on assets. Table 1 casts the position of SKAT as being consistent with a 4.4% markup over operating costs, which would imply a return on assets of only 4%. We describe the other two possible transfer pricing policies below.

Taxpayer’s Traditional Return on Assets Approach

SKAT and the taxpayer agreed that the foreign manufacturing affiliate was a contract manufacturer that did not own valuable intangible assets. The taxpayer provided a transfer pricing analysis at the request of SKAT that indicated that this affiliate purchased the raw materials and owned tangible assets used in the production of finished goods on behalf of the Danish parent. The analysis also included a benchmark study that presented the Return on Assets for several third-party manufacturers. The median Return on Assets was 9.09%, while the interquartile range was from 6.59% to 14.36%.

The Tax Tribunal decision noted the response of SKAT to this benchmarking analysis:

SKAT has then analyzed the pricing of the controlled transactions and the method used. SKAT generally does not agree with the company that ROCE can be used as a Profit Level Indicator. This is primarily due to the fact that the selected Profit Level Indicator, namely the assets, in the present case is calculated at book value. This can potentially affect the comparability between companies that have depreciated to varying degrees on their capital stock, cf. OECD Transfer Pricing Guidelines 2010, p.2.98.

Paragraph 2.98 notes that the market value (MV) of assets may diverge from the book value (BV) of assets. If this divergences impact the economic analysis, the OECD recommends an analysis based on the MV of assets. The OECD notes two of the three key reasons why the MV of assets may exceed book valuations: the MV of intangible assets exceeding book valuations and the possibility that the value of tangible fixed assets are understated by their book valuations when accounting depreciation exceeds economic depreciation. In inflationary environments, book valuations of tangible assets based on historical costs may understate replacement costs.

SKAT focused on the issue of depreciation noting that the manufacturing affiliate was recently created, so its book valuation likely captured the MV of its tangible assets. SKAT noted, however, that the third-party companies had been established for years so their accounting depreciation may have overstated economic depreciation thereby leading to a situation where the BV of assets for the alleged comparable companies were below their MV. The taxpayer presented compelling arguments that the SKAT was overstating this issue. We return to this issue after a discussion of the taxpayer’s alternative position.

Taxpayer’s Secondary Position Based on Financial Economics

The Tax Tribunal decision notes the following position of the taxpayer (complainant):

The complainant has then made an analysis of what SKAT's increases will mean in relation to the return on the company's invested capital. The calculations show that SKAT's regulation will mean that H2 will achieve a return on invested capital of between 2.0% and 4.1%. According to the complainant, this must be kept up against what an independent third party in Country Y1 will demand in return on the same investments. However, the risk premium must be adjusted for how risky the investment is expected to be in relation to the market measured via a beta value. At a beta value of 1.0, the investment in question is expected to have the same risk as the (entire) stock market. The risk premium in this situation will therefore be 6.0% (6.0% x 1.0 = 6.0%). If the investment is expected to be half as risky as the market, the beta will be 0.5, and the risk premium will instead be 3.0% (6.0% x 0.5 = 3.0%). The risk premium is thus regulated via the beta value. This also means that when SKAT assesses that H2 [...] has a lower risk than the Complainant, then H2 [...] has a lower beta value than the Complainant. In the calculations below, the complainant has identified two values for beta

The taxpayer proposed an application of the Capital Asset Pricing Model (CAPM) to estimate the appropriate Return on Assets for the manufacturer affiliate. Financial economists typically estimate the cost of capital using some version of the general arbitrage pricing theory (APT). APT distinguishes between diversifiable or unsystematic risk and non-diversifiable or systematic risk. Systematic risk is that part of an asset’s risk that is attributable to market factors that affect all firms and that cannot be eliminated through diversification. Unsystematic risk is that part of an asset's risk arising from random causes that can be eliminated through diversification. The fundamental premise of APT is that the market rewards the bearing of only systematic risk and not the bearing of risk that can be eliminated through diversification. CAPM is a variation of APT. CAPM holds that the expected return on any asset (R_j) is given by:

R_j = R_f + b_j(R_m - R_f ),

where R_f = the risk-free rate, R_m = the expected return on the market portfolio of assets, and b_j = the beta coefficient for this particular asset. The beta coefficient measures the tendency of the asset’s return to move with unexpected changes in the return on the market portfolio.

Beta coefficients are often estimated for the equity of publicly traded companies. Equity betas, which reflect both operational risk and leverage risk. Since our task is to estimate the expected Return on Assets rather than the expected return on equity, our beta estimates must be on a debt-free, unlevered, or asset basis. We shall remove the effect of leverage on these equity betas (b_e):

b_a = b_e/(1 + D/E),

where D = debt, and E = the MV of equity. To the degree that a company engages in leverage risk by financing its assets through debt instead of equity, the equity beta would tend to exceed its asset or unlevered beta (b_a).

The taxpayer asserted that the risk-free rate was 4% and the expected return on the market portfolio of assets was 6%. If b_a = 1, then the expected return on assets for the contract manufacturer would be 10%. Table 1 labels this scenario CAPM-high and notes that it implies an 11% markup over operating costs. If b_a = 0.5, then the expected return on assets for the contract manufacturer would be 7%. Table 1 labels this scenario CAPM-low and notes that it implies a 7.7% markup over operating costs.

The Tax Tribunal decision dismisses this use of CAPM towards the close of its decision:

The National Tax Tribunal does not find that the OECD Transfer Pricing Guidelines provide evidence for a methodological approach for pricing current controlled transactions of goods or services based on assessments of return requirements for financial investments. The pricing of continuously controlled transactions of goods or services must be based on a comparability analysis of similar transactions between independent companies in relation to the relevant comparability indicators, cf. OECD Transfer Pricing Guidelines 2010, p. 1.33 and 1.38.

This dismissal is unfortunate as CAPM has often been used to estimate the cost of capital for manufacturing affiliates. This decision also placed too much weight on the questionable TNMM approach as provided by SKAT.

In an October 2017 piece in Journal of International Taxation on contract and toll manufacturing , I appeal to both traditional transfer pricing analysis as well as the use of CAPM to estimate the expected Return on Assets.

Legitimate Book to Market Concerns Vs. the Questionable SKAT Approach

SKAT asserted that the alleged difference between accounting depreciation versus economic depreciation could have just suggested that the MV of assets for the third-party alleged comparable manufacturers by more than a factor of three. This assertion, however, appears to be implausible. The ultimate position of the SKAT was that it could rely on a TNMM approach measuring return on total costs (operating markup) without any consideration for differences in the asset intensity of the manufacturing affiliate versus the asset intensity of the third-party manufacturers. While this TNMM approach with no such adjustments is only reliable if the alleged comparable companies have the same asset-to-cost approach as the manufacturing affiliate being evaluated, the Tax Tribunal accepted this questionable approach.

An adjusted TNMM approach would estimate the arm’s length markup as:

(P/C)_t = (P/C)_c + r[(A/C)_t - (A/C)_c],

where P = operating profits, C = operating costs, and A = operating assets with the subscript t denoting the tested party or manufacturing affiliate and the subscript c denoting the comparable third-party manufacturers. (A/C)_t = 110% in this particular litigation, while (A/C)_c was likely near 35% for the typical third party contract manufacturer if assets are expressed in terms of BV.

SKAT also argued that the MV of operating assets for the allegedly comparable third-party manufacturers exceeded their book valuation. It is implausible that the MV was three times the BV if the only issue was the relationship between the economic cost of depreciation versus accounting depreciation. If the tested party’s asset-to-cost ratio exceeds the asset-to-cost ratio for the third-party manufacturers, then an adjustment for the economic effect of this comparability difference is required to determine a reliable arm's length operating markup using a TNMM approach. This adjustment, however, should be based on the MV of assets and not their BVs.

In an August 2000 piece in Tax Management Transfer Pricing Report, I noted a means for reliably applying either a Return on Assets approach or an adjusted TNMM approach if the third-party manufacturers are publicly traded companies. The difference between the MV of equity and the BV of equity should be added to the BV of operating assets.

Table 2 illustrates this approach by taking the taxpayer’s lower quartile, median, and upper quartile return on BV of assets as three comparable companies. Table 2 assumes each company has a BV of operating assets total $100 and its operating profits equal the observed return on the BV of assets times this $100. We further assume that the MV of assets equals the BV of assets for the comparable with the lowest reported Return on Assets, while the MV of assets is 1.5 times the BV for highest return on assets. We also assume that the MV of assets is 1.2 times the BV of assets for the median comparable company.

Table 2: Illustration of a Return on MV of Assets Approach (in Millions)

This illustration suggests that the appropriate Return on Assets would be between 7% and 8% if our assumptions with respect to the MV of assets is reasonable. The appropriate markup would be less than 10% but significantly greater than the position put forward by SKAT.

Composition of Costs and Asset Intensity Illustrated Using Benchmark Electronics

The SKAT approach not only assumes that the asset-to-cost ratio for the tested party and the allegedly comparable companies are similar, but also assumes similar composition of costs. Electronic contract manufacturing services (ECMS) companies such as Benchmark Electronics, Celestica, Flextronics International Foxconn, Jabil, and Sanmina-SCI tend to have cost structures where material costs represent 90% of total costs with labor costs being only 10% of total costs. Table 3 presents the financials of the German contract manufacturing affiliate assuming that only half of its total costs represent material costs. If material costs (MC) = labor costs (LC) = $500 million, the cost plus 10% transfer pricing policy would set sales at $1.1 billion so operating profits would be $100 million. If the ratio of working capital to material costs = 20%, while the ratio of fixed assets to labor costs = 200%, the overall operating asset (A) to total cost ratio is 1.1 and the implied Return on Assets = 9.1%. These assumptions are consistent with the court’s discussion.

Table 4: Financial Ratios for the Manufacturer Affiliate and Benchmark Electronics (in Millions)

The financials for Benchmark Electronics for the 3-year period ended December 31, 2019 were used to provide an illustration of what types of financial ratios would be provided by the use of ECMS companies as alleged comparables. Over this period, total costs averaged $2.35 billion per year and sales averaged 103.3% of total costs. Table 4 assumes that 90% of the total costs represented material costs.

The working capital for Benchmark Electronics is defined as its contract assets, inventories, and trade receivables net of trade payables. Working capital was approximately $500 million, which is just over 20% of material costs. Table 4 assumes its fixed assets represent $470 million, which is also 2x labor costs. The MV equity for Benchmark Electronics was approximately equal to its BV for this period. Net plant, property, and equipment was recorded as only $210 million, but operating lease rights-of-use assets represented another $70 million. Table 4 also includes the recorded value of goodwill, which was $190 million. Most of this recorded goodwill represented what was recorded in the purchase price allocation for the recent acquisition of Secure Communications. This recorded goodwill may economically be due to the MV of fixed assets not assigned by the purchase price allocation.

Even though the ratio of fixed assets to labor costs and the ratio of working capital to material costs for the comparable company was at least as high as the corresponding ratio for the affiliate, the very different composition of costs led to a situation where the asset-to-cost ratio for the alleged comparable was much lower than the asset-to-cost ratio for the affiliate. The use of Benchmark Electronics 3.3% markup for the markup for the tested party is not a reliable approach without some adjustment for the vast differences in the overall asset-to-total cost ratios.

This particular third-party manufacturer had a return on operating assets equal to 8%, which is somewhat lower than the 9.1% Return on Assets for the affiliate under the transfer pricing policy. A modest reduction in the 10% markup may have been warranted even though the SKAT approach was certainly a very biased application of TNMM.

Concluding Remarks

Markups for contract manufacturing affiliate should be based on their asset-to-cost ratio. The taxpayer in this particular litigation provided fairly standard evidence to justify its 10% markup and went one step further by providing at least some logic based on CAPM that its Return on Assets analysis was reasonable.

SKAT raised a legitimate question as to whether the observed return on BV of assets for the alleged comparable companies reflected a divergence between the return on BV versus MV. In my career as a transfer pricing economist, I have witnessed some taxpayers try to justify returns to capital equal to 20% or more based on inappropriate applications of this approach. It does seem questionable, however, to suggest that a Return on Assets less than 10% is several multiples of the arm’s length result.

The only alternative presented by SKAT was a questionable application TNMM to benchmark an arm's length operating markup, which ignored potential differences in the asset intensity of the manufacturing affiliate and the companies used as potential comparables. For example, the observed markups for electronic manufacturing service providers are often less than 4% precisely because their operating asset to operating costs ratios are less than 40%.

Capital adjusted markup approaches and Return on Assets approaches can yield reliable results if the analysis is based on the MV of assets for both the tested party and the alleged comparable companies. In this litigation, neither party questioned whether the BV of the affiliate’s operating assets captured their MV. SKAT raised interesting questions as to the reliability of using the BV of the comparable’s assets instead of their market valuation. If these alleged comparable companies were publicly traded, the analysis could have adjusted these book valuations to reflect market valuations.

References

Case No. SKM2018.173.LSR.

Harold McClure, “Prime Plastichem Nigeria — the Du Pont Litigation Revisited”, Tax Management International Journal, April 2020.

Harold McClure, “Misapplication of the Resale Price Method: LATAM Petroleum Distributor,” EdgarStat Blog, February 23, 2021.

Harold McClure, “Made in China, Sold by Hong Kong: Processing Trade and Transfer Pricing”, Journal of International Taxation, October 2017.

Harold McClure, "Measuring Assets for a Transfer Pricing Analysis: Book or Market”, Tax Management Transfer Pricing Report, August 9, 2000.