Inventec’s Czech Contract Manufacturing Litigation: Benchmarking Issues from an economic perspective

This is a guest post by Harold McClure, a New York City-based independent economist with 26 years of transfer pricing and valuation experience.

Several weeks ago, an LCN newsletter looked at the case of Inventec (Czech), s.r.o. a subsidiary of the Taiwanese based multinational Inventec Corporation. We thought it would be instructive to take a closer look at the benchmarking issues which this case highlighted. As alluded to in the previous newsletter, the recent Facebook and Alcoa cases also concerned benchmarking issues, but with the notable uptick in foreign direct investment (FDI) by manufacturers in Central and Eastern Europe (CEE), benchmarking disputes, especially in sectors like electronics, automotive, and energy, are likely to become more common.

It is worth refreshing ourselves on the broad outline of the dispute before we take a closer look at how benchmarking models work from an economic perspective, specifically the issue of asset intensity, which has largely been overlooked in the commentary on this case to date.

This Czech subsidiary acted as a contract manufacturer that produced server racks for Hewlett-Packard. The Czech affiliate purchased materials from its Taiwanese parent and assembled the finished goods. The cost of materials represented approximately 87.5 percent of total costs with labor costs represented 12.5 percent of total costs. As David Zářecký, Transfer Pricing Adviser at TP Tuned, summarised:

At issue was whether those costs were truly pass-through or whether they should carry a profit markup under the transfer pricing methods applied … During a routine transfer pricing audit, Czech tax authorities applied the Transactional Net Margin Method (TNMM) with a Return on Total Costs (ROTC) profit level indicator. This approach includes material costs in the markup base.

The taxpayer asserted that a better method would be Return on Value Added Costs (ROVAC) excluding material costs. The taxpayer’s approach treated material costs as a pass-through cost which did not require a markup. The taxpayer asserted that the Czech affiliate bore no economic risks with respect to the materials as this affiliate was a toll manufacturer even though they held legal title to the materials. As Paul Sutton noted in the LCN newsletter:

The court backed the tax authority’s rejection of Inventec (Czech)’s profit allocation method. Inventec used ROVAC (Return on Value Added Costs), excluding the cost of materials, arguing that it merely assembled the server racks in question and bore minimal risk. The Czech tax authority disagreed, asserting that ROTC (Return on Total Costs), which includes material costs, was more appropriate given Inventec’s formal ownership of raw materials. The judgment in the tax authority’s favour underscores a growing judicial emphasis on legal form and economic substance alignment.

The tax authority prevailed in what was a modification to its original position that both material costs and labor costs should have a markup based on the return to total costs for certain third-party manufacturers deemed to be comparable companies. The modification set the markup over material costs to 24.62 percent of the markup for labour costs on the basis that the Czech affiliate incurred less risk on material costs than it incurred on labour costs. The court decisions did not specify what these markups were or even how they would translate into a return on the assets of the Czech affiliate. The Czech courts accepted the tax authority approach on the legal grounds that the Czech affiliate retained legal ownership of inventories. As Zářecký noted, the Czech courts were asserting:

ROTC was a legitimate method because it reflected the formal accounting and economic reality. Legal ownership of the materials implied risk-bearing, regardless of what limited the company’s role was. The tax authority made a reasonable adjustment to reflect the company’s limited functions and risks, applying a reduced markup (24.62% of the standard market rate).

Economists view the appropriate markup over costs as the product of the asset intensity times the expected return to assets. While the benchmarking in this litigation appears to be addressing risk as an element of the expected return to assets, there is no mention of the asset intensity for this Czech manufacturing affiliate. Observed markups for third party electronic manufacturing services (EMS) companies are often less than 5 percent because their overall operating asset to cost ratios are less than 50 percent (this is due to the cost base being heavily weighted towards material costs and inventory to material costs, which tend to be very low due to just-in-time inventory practices).

A Toll Manufacturer Model

Ronald Simkover, the economist for the Canadian Revenue Agency (CRA) over 30 years ago, developed a position that the markup over labour costs for a toll manufacturer would substantially exceed the markup over total costs for a traditional contract manufacturer. Toll manufacturers differ from contract manufacturers in one key respect – they leave the function of sourcing components to the principal with the implication that they do not bear material costs, and they do not hold inventories. The issue is: what is an appropriate markup over labour costs? Representatives of multinationals often argue that the markup for a contract manufacturer is an appropriate metric for the markup for a toll manufacturer, but the CRA and a growing number of tax authorities worldwide disagree. The position of China’s State Taxation Administration (STA) on transfer pricing for toll manufacturing affiliates is similar that of the CRA. The STA’s position was noted in the United Nation’s Transfer Pricing Manual where paragraph 10.2.5.8 states:

Toll manufacturing is a common form used by MNEs in developing countries, but its proper return is difficult to determine since there are only a few independent listed companies that perform such activities. Some taxpayers simply use the FCMU for contract manufacturers as the markup for toll manufacturers. This grossly underestimates the return to toll manufacturers. Others use return on assets as a profit level indicator based using contract manufacturers as comparables, and this may also underestimate the return, particularly for labour intensive toll manufacturers as often being the case in developing countries.

FCMU stands for full cost plus markup, which is equivalent to the return on total costs. The notation the markup over for publicly traded contract manufacturers, which purchase components, is consistent with the Simkover hypothesis in my discussion of this issue. Simkover’s argument was that the markup over total costs (m) for a turnkey contract manufacturer should be seen as a weighted average of the return to value-added expenses and the return to pass-through costs:

m = x·v + (1 – x)z,

where x equals the ratio of value-added expenses relative to total costs, v represents the ratio of operating profits attributable to employing labour relative to labour costs, and z represents the ratio of operating profits attributable to working capital relative to pass-through costs. As an example, let the markup for a contract manufacturing that takes title to material (m) = 3 percent, while the ratio of labour costs to total costs (x) is only 12.5 percent. If the markup over labor costs = 87.27 percent and the markup over material costs = 21.82 percent, then the weighted average markup would be 3 percent. That the markup over material costs is just under one-fourth of the markup over labour costs is consistent with the benchmarking over labour costs as noted by Zářecký’s summary of the Czech court decision.

This benchmarking exercise, however, may not be consistent with the standard model of contract manufacturing based on relative asset intensities. These markups can be seen as the product of asset intensities times the appropriate return to assets with:

v = Rf(fixed assets/labor costs) and z = Rw(working capital/material costs), where
Rf = the return to fixed assets and Rw = the return to working capital.

The balance sheet information for the manufacturer affiliate would be needed to implement this model. Let’s reasonably assume that the value of fixed assets = $200 million and the value of working capital held by the contract manufacturer = $100 million. Overall operating assets would therefore be $300 million, and a 3 percent overall markup would be consistent with a 10 percent return on assets. Table 1 presents our illustrative model.

Table 1: Financials for Czech Affiliate as Contract Manufacturer (millions)

Note that we have reasonably assumed that this EMS affiliate’s ratio of fixed assets to labor costs = 160 percent while the inventory to material costs ratio is only 11.43 percent. Table 2 presents the implications of converting to a toll manufacturing structure under alternative assumptions with respect to R_w.

Table 2: Toll Manufacturer Markup Under Alternative Value for R_w

If we assume that R_w = R_f = 10 percent, then the markup over labour costs (v) = 16 percent so the conversion from contract manufacturer to toll manufacturer lowers profits from $30 million to $20 million. If we assume that R_w = 5 percent, the implied R_f = 12.5 percent. The markup over labour costs in this case would be 20 percent, which implies that the conversion from contract manufacturer to toll manufacturer lowers profits from $30 million to $25 million.

Concluding Remarks

Contract manufacturing structures have a long history in transfer pricing. In many of the North American issues between nations such as the U.S. and Canada or the contract manufacturing affiliates in China, the multinational chose a toll manufacturing structure, which led to controversies involving the appropriate markup over labour costs. As noted at the outset, the increase of FDI in the form of manufacturing affiliates in CEE countries is likely to see this issue arise more frequently in European tax disputes.

Paul Sutton rightly notes that if Inventec had wished to have a toll manufacturing structure, it should have designed its facts to not have inventories on the books of the manufacturing affiliate. David Zářecký attempted to explain what appears to be a confused benchmarking exercise in a situation where there was some debate as to whether the manufacturing affiliate deserved the return to inventories. This blog ties this all together by providing an illustration of the benchmarking issues from an economic perspective.