A draft toolkit addressing the lack of transfer pricing comparables was released for public comment in January by the Platform for Collaboration on Tax, a joint effort of the IMF, OECD, UN, and World Bank Group. The closing date for comments was extended to 7 April following the release of French and Spanish translations.
This is important work, and the consultation offers the transfer pricing community a unique opportunity to share insights about improving the way in which transfer pricing comparables are used and developing alternatives to the reliance on comparables so that practical guidance can be further refined.
The following is an adaptation of my comments on this key initiative, focusing on minimising the prominence of geographic screening and maximising the use of financial ratio screening, as well as providing some thoughts on profit splits and safe harbours.
I encourage others to submit their comments by the 7 April deadline.
Broadening the search process
The draft toolkit, in section 4.2 on page 33, suggests broadening the typical search criteria, allowing data to be used rather than unnecessarily discarded. Box 10 in that section specifically refers to potential use of comparables in foreign markets.
Use of broadened search criteria is necessary, in my view, and has already been adopted in practice. The prominence of geographic screening may lead to erroneous conclusions that there are no useful comparables, and the focus on narrow industry classification codes may serve to diminish comparability.
I am concerned that the discussion in the draft toolkit may be interpreted as expressing caution about broadening the selection criteria and may, pending the further work recommended in the draft toolkit, discourage exploration of how to make best use of available data.
Since the potential comparables resulting from a broadened approach may appear to be very different from those that may be perceived to be expected, explicit endorsement of the specifics of a broadened approach by the Platform for Collaboration on Tax is necessary to provide greater confidence that the approach can be relied upon.
Transfer pricing comparables & geography
One of the problems encountered in effectively performing a search for potential transfer pricing comparables using a commercial database is largely self-inflicted, caused by the elevation of geographical market in the order of screening criteria (as Box 7 on page 28 shows and seemingly supports, geography is typically the second screening criterion).
The elevation of geographical market immediately reduces the number of potential comparables simply because of data limitations. As Appendix 4 tellingly summarises, there are over 150 countries that are unlikely to have sufficient data to provide comparables relating to their country, yet there are tens of thousands of data points for over 30 countries.
Elevating geography as a screening criteria tends to emphasise the paucity of data for particular countries, ignores the wealth of data that could be utilised, encourages a perception that geographic comparability overrides other comparability factors, and perpetuates the perception that useful comparables do not exist. In turn, this puts pressure on the arm’s length principle.
The typical elevation of geographic comparability over other comparability factors as a screening factor is also difficult to justify on technical grounds. It seems more likely that the use of geography as an initial screening factor has developed as a practice for dealing with large volumes of potential data points in those countries that are data-rich.
In any event, there are many comparability factors for the potential comparables about which very little or nothing is known (for example, contractual terms, components of pricing, risk profile, business strategy, product or services mix, markets of customers), yet we are content to assert approximate comparability. It may seem odd that the one factor we do know about, country of incorporation, is treated more uncompromisingly.
Further focus on geographic comparability will arise when tax administrations have access to country-by-country reporting data and can compute and consider the margins reported by an MNE around the world for apparently the same routine functions.
If geographic comparability continues to be elevated in the screening process, there may seem to be something wrong about an MNE reporting consistent rates of return for similar low-risk activities around the globe, and tax jurisdictions that seem to be systematically disadvantaged through the vagaries of geographic screening may turn to different approaches. Both outcomes would be unfortunate.
My view is that, even without further research work being undertaken, it is not sensible to elevate geography as a screening factor above other comparability factors, particularly for those countries included in those 150 or so listed in Appendix 4 that will have very few data points anyway.
Therefore, I suggest that consideration is given to provide guidance in the toolkit that initial geographic screening in a typical search process is useful only when the tested party operates in a country for which there is an abundance of relevant data points that provide an adequate sample for further refinement.
If there were something important for comparability about the particular country, then this might be adjusted for in the comparable data derived from other geographies, as the draft toolkit indicates.
This seems to be a better approach than continuing to use a geographical screening criterion but extending it to other countries in the region on the dubious assumptions that countries are comparable since they are near each other, and that countries far away are not comparable.
The toolkit should also specifically state that the impact of potential country features on the rate of return for low-risk activities is likely to be limited.
When the local function is relatively low-risk, and therefore somewhat insulated from market conditions, and with limited exposure to its working capital (which may itself be limited), it is less likely that the local function needs to be adjusted for country risk factors that are effectively assumed by the other party to the transaction.
Economically significant financial ratios
Moreover, I suggest that consideration is given to providing guidance in the toolkit that economically significant financial ratios be elevated in the typical screening criteria.
Box 10 hints at this when it refers to finding potential transfer pricing comparables in other geographies and in other sectors, and the guidance should be developed to encourage such an approach.
In addition, I suggest that any further work on reliability of transfer pricing comparables from other geographic markets should also consider the impact of potentially greater comparability provided by alignment of economically significant financial ratios.
In other words, the work should not just examine, for example, manufacturing returns by industry classification across different countries, but should look at manufacturers in different countries with similar ratios, such as fixed assets to total costs or employee costs to total costs.
Further research should not focus on whether the search process starts with homogeneity in rates of return for activities with the same industry classification across countries, but on whether the process, after making comparability adjustments (including alignment of economically significant financial ratios), ends with homogeneity for parties in different markets.
Economically significant financial ratios are much the same concept as diagnostic ratios, discussed in section 6.3.1 of the toolkit, but the stage at which they are used is different.
Diagnostic ratios are very useful in identifying the more comparable enterprises in the final set. However, they are also useful as a screening criterion early in the process of selecting transfer pricing comparables and before other criteria are used which may unnecessarily limit the data available or which may screen out potentially useful comparables.
Concern about the elevation of geography in the screening criteria can extend to the elevation of industrial classification codes.
Where there is a large amount of data to filter, such codes can be useful. But where there is a small amount of data, such codes, if used narrowly to focus on product rather than on activity, can remove potentially useful comparables.
This is particularly the case when the search criteria are not aligned with the accurate delineation of the actual transaction, which, as the draft toolkit correctly emphasises, is often much more important than the mechanics of benchmarking.
For example, the accurate delineation of the transaction involving a controlled pharmaceutical distributor may show that it markets a small range of specialised drugs, outsources box-shifting to a third-party logistics provider, assumes no inventory or credit risks, and employs a large sales force. Turning to a database and beginning with an industry classification filter relating to a pharmaceutical goods importer (wholesale), though, will not likely retain alignment with the accurate delineation of the controlled transaction.
If the resulting supposed potential comparables are those of aspirin-to-Zimmer wholesalers with large warehouses and distribution infrastructures, significant inventory and credit risks, and a relatively small sales force, then probably the only thing in common with the tested party is a shared industry classification.
The economics of the two businesses are far from comparable. Applying diagnostic ratios to such an inappropriately filtered set to find potential comparables with, say, lower fixed asset ratios, lower inventory ratios, and employee to cost ratios aligned with the tested party, is like looking for a banana in a box of apples, and pretending the most yellow apples are bananas.
Starting with a box of bananas is better. The accurate delineation of the transaction suggests that the activity is that of marketing, and rather than try to turn an aspirin-to-Zimmer wholesaler into a marketer, it may be more useful to start with marketing activities across all products and to screen at the initial stage for significant financial ratios (for example, the tested party may have a low ratio of assets to total costs, and a high ratio of employee costs to total costs, both of which may be relevant screening ratios when searching for comparable marketing activities).
I suggest, therefore, that consideration is given to providing guidance in the toolkit that financial screening ratios can be used to filter data to achieve greater comparability with the tested party than that achieved by narrow industry classification codes which may, in practice, diminish comparability.
Order of screening criteria
It would be helpful if the Platform for Collaboration on Tax could be very clear on the ordering of screening criteria, and in particular whether it is necessary to elevate geographic screening or use narrow industry classification codes, since the results from a broadened approach will look very different to some current expectations of typical benchmarking.
For example, instead of wholesalers of after-market car parts being used as apparent comparables for national marketing subsidiaries of global car manufacturers, one may propose more reliably under a broadened approach a set of comparables that include marketing activities involving mobile telecommunications products or kitchen appliances that show relevance to the economics of the delineated functions, assets, and risks.
Instead of computer assemblers being used as apparent comparables for chip manufacturers, one may propose more reliably under a broadened approach a set of comparables that include food manufacturing and which demonstrate similar key financial ratios. Instead of comparables from the same country or region, one may propose more reliably relevant comparables from other countries.
The direction suggested by Box 10 is that comparables for the distributor of mining machinery in Country A could be a distributor of building materials in Countries X, Y, and Z. All these scenarios are not only possible to envisage, but are being explored and adopted in practice so as to make the best use of data.
It is thus not helpful if readers interpret the draft toolkit as not endorsing such approaches, and it would be a considerable improvement if the draft toolkit made specific recommendations about flexibly substituting broader functional comparability and financial screening ratios for narrow industry classifications and geographical market screening.
To utilise the wealth of potential comparables available in this manner and strengthen application of the arm’s length principle, specific endorsement from the Platform for Collaboration on Tax will be valuable.
The discussion of profit splits in Section 5 of Part III is likely to touch on some sensitive and strongly-held views about whether, and if so to what extent, profit splits open a valve to release pressure on the arm’s length principle based on comparability.
On the one hand, profit splits, coupled with valuation techniques, can be used to value royalties in the absence of comparables, but on the other hand, profit splits are a highly specialised method that can be reliably applied only in very specific arrangements. As the draft toolkit indicates, a lack of comparables alone is insufficient to warrant the use of a profit split under the arm’s length principle.
Nevertheless, a profit split does have a potential advantage, at least in some types of application, in that it starts with an actual profit rather than a benchmarked profit; although it is true that the profit on the transaction accruing to the parties may not be readily computed.
Also, when it is difficult to bring reliable comparables to bear, experience suggests that practitioners and tax officials do try to evaluate a sharing of total profits, taking into account the contributions of the parties as best they can.
However, the sensitivity arises because this is not an application of the profit split as set out in chapter II of the OCED transfer pricing guidelines; the bar determining whether there are reliable comparables is not capable of objective measurement and may be set too low, and there are no ground rules about what principles should apply in evaluating the appropriateness of each party’s share. There is likely to be additional focus on the evaluation of how profit is shared when tax administrations review country-by-country reporting data.
The appetite for using some kind of profit allocation approach is not likely to diminish, and so the development of ground rules would be helpful while maintaining a distinction between such an approach and the profit split method described in chapter II of the OECD guidelines. The following paragraphs offer some suggestions that might prove stimulating in considering potential guidance for using profit allocations in the absence of transfer pricing comparables.
One way of verifying the allocation of profit in the absence of comparables might be termed a comparable allocation method. (Seeds of this approach may be seen to be planted in some aspects of the discussion of profit splits in chapter II of the OECD guidelines, but they have not germinated.)
For example, suppose there are reliable comparables for an MNE’s controlled transactions in the European market and that these determine the arm’s length profit for the European distributors and production entities; suppose also that these benchmarked profits result in an allocation of total profit for the European market of 35% to distribution, 25% to manufacturing, and the balance to services and intangibles contributed by the global head office.
In the absence of comparables for the distribution and production entities in Latin America, for example, or to support the reliability of potentially inaccurate comparables, the allocation determined for Europe could be applied to the total profit for the LatAm market.
In this scenario, it is assumed that the same functions are performed by the distribution and manufacturing entities in both regions, or that accurate adjustments to the allocation of profits could be made for any differences in functions, and that the functions are entrepreneurial and involve the assumption of risk. If the functions were low risk, it is less likely that reasonably accurate benchmarking of the LatAm entitles could not be performed, or that their results would be significantly influenced by potential differences in local market conditions.
Another way might be to draw the not unreasonable conclusion that a controlled transaction with no open market reference should be regarded as an internal arrangement.
There are ground rules for evaluating internal arrangements set out in the 2008 and 2010 OECD reports on the attribution of profits to permanent establishments.
Under this suggested approach, one would deem a transaction between the associated enterprises for which comparables cannot be found to be conducted for the purposes of evaluation within a single enterprise. There would still be no comparables available to price the “dealing,” but the focus on the location of significant people functions and the attribution of risks and assets may prove a useful framework for attributing the shares of the combined profit to the parties.
Sanity checks on profit allocations that are currently used are probably closer to an attribution of profit approach than to a formal profit split as set out in chapter II of the OECD transfer pricing guidelines, and so acknowledging the approach would at least introduce some agreed rules.
If for no other reason, the complexities associated with deeming a transaction an internal arrangement in such cases may serve to encourage a more intensive examination of what might be available to use as comparables.
Transfer pricing safe harbours
Transfer pricing safe harbours can in principle reduce the reliance on benchmarking studies and can be designed to align with a country’s fiscal policy objectives and sources of economic activity.
However, eligibility criteria can be difficult to determine, and tax administrations may tend to design narrow entry definitions that can reduce their effectiveness. As an aside, descriptive eligibility criteria can be difficult to draft and interpret; instead, just like the use of economically significant financial ratios in the context of screening for comparability, financial ratios could also be used to determine eligibility appropriately and clearly (for example, marketing spend above a certain ratio to total costs would lead to non-eligibility).
Nevertheless, improvements to eligibility criteria may not improve the effectiveness of the safe harbour so much as a safe harbour which is more universal in design, but with specific exemptions. Such an approach, for example as outlined below, may be feasible.
Risk & safe harbours
It is noticeable that the examples of transactions for which safe harbours may be appropriate (page 55) all have in common the absence of significant risk. In turn, the absence of significant risk means that the expected profitability across a wide range of activities is likely to be predictable and stable within a narrow range, and may be less affected by any market differences between countries, leading to greater prospects of the adoption of common approaches across countries.
It may be worthwhile considering harnessing this common feature to provide guidance on designing an effective safe harbour that covers all intra-group activities where control of economically significant risks is not exercised in the country, with the exception of certain specified sectors which may be important for the country’s economy or which present other policy concerns or risks factors (e.g., mining, software, insurance, as the case may be).
This safe harbour would likely require the taxpayer to provide supporting material, which could include contract terms and a risk statement modelled on the principles of chapter I of the OECD guidelines. One could spend resources on defining appropriate profit level indicators and various profit levels, but such an approach brings complexity.
If one of the key reasons for turning to safe harbours is to avoid the difficult task of finding reliable comparables, or indeed any comparables, it may be sensible to adopt one-size-fits-all: that is, one profit level indicator (I suggest mark-up on total costs), and a fixed mark-up rate.