UT parameters – Sales

“Sales” is the amounts received from selling anything outside the Business, grouped by the residence of the customer rather than the supplier.

Picciotto refers to the Business needing to have an establishment in the country of the customer, so it can be said that it is doing business there, but this is a much broader term than your normal Article 5 PE under current rules.  His paper is silent about sales to a customer in a country in which you have no establishment.

I was regarding the PE issue as just one aspect of the Sales factor, but it seems to be significant enough to deserve a post of  its own, so I’ve hived it off.

Inter-Business sales

Sales within a Business are ignored, but those to another group company outside the Business seem to count.  Where sales are cross-border, this would seem to raise the spectre of transfer (mis)pricing again.  One of my aims here is to eliminate transfer pricing as a consideration, but this is giving me some problems.  My approach with the definition of a Business was to say that there should be a threshold above which the level of transactions is sufficient to bring the two activities into the same Business, and below which the amounts are going to be insignificant and so are taken as booked.

I’m not entirely happy with this approach, but I can’t see a sensible alternative.  You either eliminate all the transactions by making the entire group into a single Business, or you allow multiple businesses but have to find some way of pricing related party transactions.  If we can do the latter properly then we might as well not bother with UT at all; if we take it as a UT axiom that this is not possible, sacrificing the idea of separate Businesses is the only way to go.  On the other hand like to preserve them where possible – if only to allow different formulary apportionments for different industries – so I’m going to compromise by using the Business split method above.

Recording sales

Where we have a commission-based model, what counts as sales?  Take a UK software reseller, dealing in products from a US company.  If we have a 10% commission, then you can look at this two ways: either the UK sales are 100 and cost of sales (which is disregarded to UT) is 90, or UK sales are 10.  The accounting treatment of this could dramatically affect the Sales factor in the UT formula.  For the moment I think I need to just stick with the accounting treatment and trust the accountants to come up with the right answer.

We also have a mismatch between groups and separate entities.  If the reseller were part of the same US Business that produced the software, then the whole 100 would be recorded as UK sales; if the reseller is a separate Business from the developer then even if we agree that only 10 is to be recorded as reseller’s sales, so with the 90 recorded by the developer all 100 is in the UK, the 90 is pulling in very different profits.  It’s hard to see the impact of this before I model it, but it’s a potential asymmetry that could cause a problem.

B2B transactions – goods

I’m not at all sure how one can track the destination of a sale by one global entity to another.  Say a US manufacturer sells goods to the UK-based purchasing company of an entirely unconnected retailer (via that company’s US branch office), which then sells them in the UK, France and Germany.  Where does the manufacturer record the sales?

1)       all in the UK

2)      all in the US, or

3)       in the UK, France and Germany?

I think (3) is unacceptable: it would rely on the retailer passing on its sales information, which is not only a commercial issue but would mean that the manufacturer is reliant on a third party to determine its tax position.  (2) seems odd, if none of the goods are being sold in the US, but then (1) seems odd as the manufacturer doesn’t really care where the goods end up if all the talking has been done in the US.

I ws hoping the CCCTB principles would help, but they are more about assigning sales to the right entity, rather than working out which country the sale should be allocated to.

The answer presumably lies in movement of goods, and VAT rules are probably helpful here: broadly, the supply of goods is where legal title to them changes.  So if the retailer takes control of the goods in the US and arranges shipping to the UK itself, then the sales are in the US.  If the manufacturer ships to the UK and the retailer takes over from there, then sales are in the UK.  And if the manufacturer ships to France, then the sales are in France.  This seems like a fairly simple rule to apply, although I can see it might be easy to game.

Of course we’re also in the slightly odd situation that when a US company signs a contract in the US with a UK company, this is a sale made in France.  If we regard this as the US manufacturer profiting from the French market for its goods then that seems reasonable, but that does rely on equating the shipping chain with the source of profit.  What if all the goods are shipped to Rotterdam, the retailer takes control there, and they are then sold in France and Germany?  If we call that Dutch sales then the Netherlands make something of a windfall gain.

This also seems to tie in very much to PE issues, as of course one UT axiom is that you can only have sales in a country in which you have a PE.  If our US manufacturer has no presence in Europe then things are going to be much simpler in one respect, but see my thoughts on PE.

A thought occurs: what if the shipping contract is such that the manufacturer has title to the goods when they’re loaded, then hands them over mid-Atlantic and the retailer is responsible for unloading?  It doesn’t happen now, so far as I’m aware – but if it had a major tax impact…?

B2B transactions – services

This seems trickier even than goods.  If a US consultancy sells advice to the European retailing group above, where does it do so?  We can’t follow anything tangible.  Possibilities:

– Where the staff providing the services are.  This kind of ignores the whole point of UT.
– Where the providing company is resident.  Even worse.
– Where the purchaser makes use of the services.  This is a bit vague and hard to define, and it may be hard for the taxpayer to get any reliable information.
– Where the purchaser under the contract is based.  This is clear, but means that the sales get concentrated into one place even if they’re global.  It would also seem to let a business purchaser dictate where the provider pays tax, which may permit abuse.

I think the only coherent answer is to look at the contracting entity, and then maybe see if any safeguards are needed.

Initial conclusions

Follow the accounting recognition of sales.

Intra-Business sales are ignored.  Sales between related businesses are recorded at the accounting values, which should be reasonable.

Sales of goods are allocated to the country in which legal title to the product changes hands.

Sales of services are allocated to the country in which the purchaser under the contract is resident.

See the PE post for what to do if sales are made to a country in which there is no PE.


This is an area that is fundamental to UT, but seems very hard to pin down once you start looking at it in any detail – especially on the B2B side, and of course one of the points of UT is to address the taxation of large businesses.

I note that Clausing and Avi-Yonah in 2007 suggested that using VAT principles would make life easy, but then didn’t follow that through.   I haven’t found anything else that goes much further than saying “allocate sales by destination”.  If anyone knows of any further work in this area, it’d be good to have a look at it.


UT parameters – Labour

“Labour” is defined as a combination of payroll costs and headcount; they are weighted 50:50.

It is the place of work that counts, not the residence of the employer.


I am going to assume that we take the average number of employees on the books through the year.  I’m not sure whether to count noses, or to count full-time equivalents.  Following accounting practice I will go with noses, but there are a couple of areas that might need to be addressed:

  • Two people on 16 hours a week will weight the formula more than one on 35.  This may be either a bug or a feature.
  • Should there be a de minimis threshold, as we used to have in UK GAAP?
  • Zero-hours contracts are big news at the moment: a large number of employees with brief periods of work will bump the numbers up considerably, if the Companies Act definition whereby any work in a month gets you 1/12 of a headcount is adopted.

We include all persons working for the firm, including employees of subcontractors providing labour services.  I assume this means staffing agencies rather than all sub-contractors, but there is a bit of difficulty drawing a bright line between provision of staff and provision of services.  Looking at various classes of worker I find my view is rather coloured by UK employment/self-employment tests:

  • Engaged personally under a contract of service: clearly included.
  • Engaged through an agency under a contract of service: defined as included.
  • Engaged as self-employed: probably excluded
  • Engaged on a piece work basis: included if controlled by the business, but excluded if link is loose
  • Staff of a firm providing professional services: excluded.
  • Former staff outsourced en masse to a third party: I’m tempted to include them, but what happens after five years when all the original staff have left?

For clarity I’m going to take a simple definition and only include employees with a solid contract of service, even if it’s through an agency.  If there is any doubt, then they are excluded.  I appreciate that this may permit outsourcing to reduce headcount “artificially”, but console myself that the outsourcing company will include them in its own UT calculations and they will still pull profit to the country.

An alternative measure would be to look at FTEs, perhaps by taking hours worked and dividing by whatever a full-time week is.  Although as all we want is a relative number the raw hours worked would do perfectly well, with perhaps an assumption of 40 hours for salaried staff with no set hours.  But we then get into problems with piece work, and whether directors count as employees, and record-keeping generally.  And should workaholics be capped to 40 hours even if they work 70?

Payroll costs

Clearly we need to include wages and salaries.  Bonuses would seem to be fair game.  Benefits should also be included, although share-based remuneration gives me some pause for thought: I think however that this is just my historic uncertainty over why we take the cost to the P&L in the first place rather than treating it as a cost of the shareholders (I’ve never been philosophically convinced of that one).  Pensions and SAYE and so on should be included too.

Staff welfare generally is an odd one, especially as it is hard to draw a bright line in the spectrum running from operating costs through staff welfare to benefits.  I think the touchstone should be along the lines of “would it be taxable on the employee?”, although given the different tax and benefits regimes we would need to define this more closely.  For my purposes, then, the costs include any amounts which in the UK would go on a P60 or P11d, or be included in a P11d dispensation or PSA.  Amounts which are specifically exempted from that reporting requirement are also included: I want the underlying principle, not the actual taxable amount.

For subcontractors, the actual staff cost is not going to be the amount paid to the subcontractor, as that will include an element of profit.  As however that profit should hopefully be taxed in the same jurisdiction by virtue of the sub-contractor’s staff being there, I don’t propose to adjust for that.

Should social security borne by the employer be included?  I think not, as this would be slightly distorting: by increasing the secondary NI rate, for example, the UK would not only rake in more NI but would also increase the company’s allocated profit: the Chancellor may then be tempted to hike secondary NI and cut CT slightly.  I am slightly uncomfortable at excluding a cost so closely tied in with employment, though, and recognise that this will further increase the impact of outsourcing and using staff suppliers (as the business will then effectively have to include the social security cost inherent in the agency fees).

Initial conclusions

“Labour” is people engaged on a contract of service, whether directly or indirectly.

Headcount is the average number of those people over the year.  No account is taken of working hours.

Payroll cost is the total of amounts paid to people included in the headcount, including benefits and any amounts that could reasonably be taxed on them; plus the amounts paid to sub-contractors in respect of labour (but not services) supplied.

Social security and other costs borne by the employer but which the worker does not benefit from are excluded: only amounts paid to the worker (or agency) count.

I am uncomfortable about a number of aspects of this, as I have had to draw some very simple lines and exclude some costs which are economically very similar to those included.  My gut feeling is that there is going to be scope for a lot of gaming of this area.

UT parameters – Assets

Apologies for the delay: it’s been a busy couple of weeks.

The first element of the UT allocation formula is Assets.  This includes all fixed tangible property, including leased assets (to avoid sale-and-leaseback avoidance).  It excludes inventory (and so by extension all other circulating capital?).  It excludes intangibles, because it’s hard to tell where they are and they can move about easily – we look through the intangibles to the assets and staff that create them, and anyway we are more interested in how the IP is used than how it is created.

Taking normal tangible fixed assets at first, the key question is how to value them.

The other elements of the UT formula, Sales and Labour, appear to be on a current-year basis, so for consistency assets should also be.   That implies that one should bring in a value which reflects the value of the asset in the current year.  I am trying to eliminate any subjectivity in the formula, which implies that I need to get an objectively-correct figure that is not open to debate.  However, the only number that one can definitely and unambiguously fix on an asset is historic cost, but this is clearly not going to be appropriate in a lot of cases – old buildings, for example.

The obvious alternatives to cost would be market value, or book value.  To use market value would mean bringing in a subjective element, and would require a lot of effort each year.  Book value also has subjective elements, in the choice of depreciation rate, and where assets are revalued.

I think that the only practical option is to go with market value in principle, but allow for approximations to be used.  So for most plant and machinery the NBV is often regarded as an approximation to MV (though in principle it is clearly not meant to be one), and so I will allow NBV to be used as a proxy for MV.

This introduces a potential gap in the system, where assets can be revalued up or down to manipulate the formula, but as the alternative is to have historic cost distort the results I think I have to live with that.

Leased assets: how do we value a lease?  As I understand it GAAP is moving to a position where a lessee under an operating lease recognises the value of the asset on the balance sheet, but brings in a matching creditor for the rent payable.  This would seem to solve that problem neatly, and the asset value would relate directly to a real-world cost.

For short-term hire, I think this has to be omitted if it doesn’t show up on the balance sheet.  I can think of a few areas where this might lead to anomalies, but those can be addressed once I have a working system and can identify them properly.

Lessors: the new accounting principles, as I understand them, require the lessor under an operating lease to remove the asset from its balance sheet and replace it with the receivable.  This would seem to distort the formula for leasing companies rather significantly.  Subject to people telling me what a bad idea this is, I will regard this sort of lease receivable as an “asset” for UT purposes, as it represents the value of something tangible, even though it is itself intangible.  I can see problems already if the lessee takes the asset across borders – perhaps the lessor should attribute assets by physical location of the asset, so an asset you let out becomes a source of income in the jurisdiction in which it is located.

I am going to use the value of assets at the balance sheet date, as that fits with accounting practice.  Ideally we would have something like an average value over the year, but that is likely to be inordinately difficult in practice.  This does introduce potential distortions, such as bringing forward the acquisition date of property to 31 December  in order to bulk up the balance sheet, but I can’t see a practical way round that.  Although I suppose one could take the average of opening and closing assets, which would mean that any movements have only half the impact (good or bad), so the effect of any manipulation would be dampened.

There is a question over cut-off: I don’t think capital commitments should count, as the assets are clearly not yet in use, but then having a test about whether things are in use or not adds an awful lot of complication.  Again, I think I have to follow accounting practice and say that if the accountants show it as an asset then it’s an asset, and then I can look at possible weaknesses later.

Where an asset ceases to be used and is slated for disposal this must be recognised, so it can be counted as stock and disregarded.

The digital economy is a tricky one.  Is a website a tangible asset or an intangible one?  What about the cloud, where we lease server time?  Paying for servers actually used is functionally the same as paying for servers we then use, but would never show up on a balance sheet.  What if the services are packaged, so we pay a single fee for website development and maintenance, hosting, storage, SaaS, etc – how do we disentangle the asset from the service, and the intangible from the tangible?  Much as I hate to hand-wave away the current leading edge of service provision, I think that this is all going to have to be classed as not a tangible asset (too much of it is intangible and/or revenue) and so left out of account.  This does lead to a possible mismatch with sales: a website is a presence in a country and so is fundamental to Sales but irrelevant for Assets.

Initial conclusions

Take the market value of tangible assets (or a measure which reasonably approximates it) at the year end date.

Assets for resale must be excluded.  Where assets are no longer to be used they must therefore be transferred to stock.

Lessees must rely on lease accounting to bring in exactly what we want to count as an asset.  Short-term operating lesaes are to be excluded.

Lessors should continue to include the amount representing their income stream from an asset as being that asset.  The location of the asset must be determined, so the amount can be attributed to the correct jurisdiction.

UT parameters – the Unitary Business

A Unitary Business is the basic taxable unit.  It consists of all legal entities under common control which operate in the same or related business.   Note that I am using “business” to mean economic activity in general, and “Business” as a technical term for this basic taxable unit.

“Control” is defined using a 50% test, but this is to be extended to catch avoidance situations.

There is a presumption (rebuttable?) that businesses are related if there are a significant number of transfers between them, or share common resources/services.

Number of Businesses

Starting with a business that buys and sells widgets, where all entities are 100% subsidiaries, this is clearly a Business.

If we also buy and sell grommets, with broadly similar suppliers, customers, assets and staff, this would appear to be the same Business.

If we have a retail operation in the UK, and another in France (with a separate brand name, staff, assets and customer base), are these separate Businesses?  If we set up an EU procurement company to supply both of them, do we now have 1, 2, or 3 Businesses?  I assume that we consolidate into as few Businesses as we reasonably can, and so would just have the one.

Similarly: if we have a widget-selling business, and then acquire a separate grommet-selling business with its own staff, premises, suppliers and customers, is this immediately subsumed into the same Business (as the businesses are related), or do we need to wait until there is some economic integration?  For example, Alliance Boots – there is a retail business which is primarily in the UK and a wholesale business which is primarily not, from which one can fairly safely conclude that there will be a lot of labour and assets which are not at all shared; but would the shared elements be enough to bring them into a single Business?

If we make & sell widgets, and buy & sell grommets, do we have a single Business or two Businesses?  Or three, indeed – one of manufacturing and one/two of selling?  Again, I assume we would combine these into one Business if we can: there is a single Business of selling, and the manufacturing is just another way of sourcing the product of one of the businesses within it.

Similarly, if we buy and sell both widgets and grommets, and have a 40% stake in the factory we buy the widgets from but get grommets from a third party, how many Businesses do we have?  I assume that we have one, and we exclude the factory as not being controlled.  Would this change if all or nearly all the factory’s sales are to us – would that be enough de facto control and integration to bring it into the Business?  Can the factory dip into and out of the Business depending on how many thrird-party sales it makes?

Shareholdings: do we look through guarantee companies and partnerships?  I assume so – the test would not simply be a shareholding test like the normal UK CT grouping test, but would look at actual control.

On that note: who exercises control, if we are ignoring companies?  We need to find some way to identify management bodies which control each Business.  This needs to be below the board of the parent company – clearly this will control all the Businesses, but if we only look at that level we could only ever have one Business.

Accounting: assume a 100% group with 3 Businesses.  How do we establish the profit of each Business if they have a shared service centre?  Or does the existence of the shared service centre itself make them into a single Business?  It would seem odd if any inter-company transaction at all made businesses into a single Business, but if there is a threshold then we need to be able to deal with transactions beneath it.  The obvious route would be  to treat the actual amount received as a Sale, and the amount paid as a nothing (unless it can be brought into the Asset or Labour pools) – but how do we ensure that the Sale value is correct if we are trying to eliminate the arm’s-length principle in favour of a purely mechanical approach?  What if the transaction is intra-company: the salary of the MD of a company involved in two Business, for example?

We have a widget seller based in the UK.  It opens a branch in France – part of the same Business.  It opens a subsidiary in Germany – same Business.  When it starts a 50% JV with a local entrepreneur in Spain – is this the same Business, if something (operational factors?) gives us control?  What if we have 40% but operational control (so it’s in our Business), and the Spanish partner has 60% (so it’s also in their Business)?  Do we need a tie breaker for dual control situations?  The simple answer must be that when in looking at the extension of “Control” to shareholdings below 50%, you can only do so if no-one else has control, but this may take some careful wording.

Where do we start with determining whether businesses are related?  We can’t start with a company, as it could be involved in several Businesses, and to start with a business seems a bit chicken-and-egg.  Do we just start by breaking down the group’s activities into some rough splits and then see how the detail fits together?  Looking at the group that manufactures widgets, buys in grommets, and sells them both, should we start small and build up if they seem linked, or should we start assuming one Business and only split them if we can’t justify keeping them together?  The latter seems simpler, but might lead to a presumption of unity which could distort things.  Starting with the former might mean we end up too fragmentary.

Initial conclusions

I shall assume that all businesses in a group are a single Business to start with, but then actively examine the activities to see if one can justify splitting anything off.

Control will be defined as the power to have the business act in accordance with the wishes of a certain agency.  I need to define the agency which can exercise this control.

I also need to define the form of control.  Shareholding is a useful initial test, but obviously doesn’t work in all cases (such as a JV controlled by a 40% shareholder).  I want something more rigorous than an elephant test; it needs to be capable of breaking ties.

I need to establish a threshold below which intra-group transactions are insufficient to form a Business.

I am trying to abandon the arm’s length principle, so inter-business transactions below the threshold will (for the moment) be left at the amount in the books and not adjusted.


I have assumed for the moment that there will be a single UT formula to apply to all Businesses.  Should there be different formulas then Iwill need to make sure that each Business only contains businesses to which a single formula shuold apply.  This may then require that the interaction threshold be disregarded, and two Businesses kept separate no matter how great their interactions.  This seems like a complication to be layered on after a single-formula system has been established.

Unitary taxation – initial questions

Trying to put together a test system, I find that it is trivial to put together one for a very simple case where one looks only at high-level figures. Once it gets more complex, however, there are many areas where there is a choice to be made about how to treat certain situations.

I don’t want to build a straw man of a system, so I’d like to flag up some of these areas to get comments to enable me to build a more robust system.

These are just initial questions, and I haven’t yet come up with answers to many of them.

I’m going to break it down into four main areas, each with their own blog post:

1) Defining the Unitary Business
2) Identifying Assets
3) Identifying Labour
4) Allocating Sales

All thoughts welcome.

Are pension funds necessarily tax avoidance?

I have been having a slightly heated debate with Richard Murphy on his blog.  Or rather, not on his blog, as he has deleted the salient comments.   He says the question is “neoliberal trolling”; I think it is a valid question.

To recap: 

Talking about Alliance Boots and the fact that there is a tax deduction for financing costs, Tim Worstall claimed that the interest would be taxed in the recipient’s hands: the implication being that there is therefore no avoidance.  That is, I think, the key issue under discussion.  Murphy replied that the interest would not be taxed as it might be moved offshore. 

“Portemat” said that in fact some Alliance Boots debt is held by UK pensions, to which Murphy replied that pension funds do not pay tax – the clear implication being that if tax is not paid then there is still a problem regardless of the reason.

I pointed out that that this is by design and so is hard to see as avoidance, to which Murphy then brought up arbitrage.  I have no idea why he considers that to be related to the issue under discussion (pension schemes not being taxable).  He then deleted my comment that I was talking about pensions not arbitrage, and has since then been a bit selective about what he lets through.

So essentially Murphy seems to have derailed a discussion because he has realised that his line of argument was not going anywhere.  He then deleted my comments, and started bandying round terms such as trite, nonsense, and demeaning, rather than address the previous thrust of the discussion.  To my mind this does not make him look as if he has a strong coherent argument.

My simple question I put to him, which he has refused to answer, was:

Going back to the core point of the discussion, then: if a pension fund lends money to a business, the business gets a tax deduction for the interest but the pension fund pays no tax on it.  Put simply: I consider this to be a reasonable position which is in line with Parliament’s intentions.  Do you?

I don’t know why he won’t answer this question.  I can only conclude that it is because if he says yes then he is a) agreeing with me and b) suggesting that maybe Boots isn’t so bad after all, and thus contradicting himself; and if he says no then he is adopting a rather extreme view that few people would agree with.  So he’s written himself into a corner.

I’m posting this mostly because I’m a bit peeved that he calls me a troll for asking a simple question, but refuses to engage in any sort of meaningful debate, so I’d like to call him on that.  He is of course not obliged to join any debates (or leave them visible) on his own blog, but then if he is going to be territorial about blog posts then so can I be 🙂

Also: if anyone can tell me what “neoliberal” means, I’d be interested to know what my political views apparently are.  I understand there is some sort of conspiracy involved, but no-one’s invited me to any meetings…

Unitary Taxation – building a test system

I have reservations about Unitary Taxation, so I’d like to build a model of how it works so I can test it for weaknesses, and compare any weaknesses which come up to the arm’s length principle to see whether UT is a better system.

At this stage I’m looking more at the mechanics of how it will work, rather than any ideological questions of whether the results are better – though I think some of those will come up along the way.

To do this I’m working mostly from Sol Picciotto’s paper “Towards Unitary Taxation of Transnational Corporations”, as this seems to be the most detailed exposition and is used as an example by Richard Murphy, Prem Sikka, the Tax Justice Network, and others.

It’s a fairly long paper, so I’ve attempted to summarise it below to give me some parameters for the system.  In my next post, which at the moment looks like it’s going to be rather longer than this one, I’ll list out some questions of detail which have arisen.  But if anyone thinks I’m misunderstanding anything so far, please do shout before I get too far along and make a fool of myself.


A UT system can be implemented unilaterally.  We don’t need agreement between states as to how it will work, we just need a reasonable amount of co-ordination with some mutual agreement procedures.

A big advantage of UT for governments is that they can’t offer tax incentives to companies to invest in their country.  Or rather: it removes the temptation to offer incentives, which is good because doing so is not generally to their advantage.  They can therefore compete on tax rate, but nothing else.

In the absence of differences in tax rates between countries in which a given TNC has economic activity, universal adoption of UT would leave no scope for avoidance.  I’m not sure how this squares with the suggestion that countries may compete on tax rates, but that is perhaps the sort of philosophical question I’m not looking at just yet.


The basic tax return for a business is a report consisting of:

–       The business’s consolidated accounts, with all internal transactions eliminated
–       Inputs and outputs of the UT formula

A UT Business is all legal entities under common control (using a 50%test , but with (unspecified) anti-avoidance rules) which operate in the same or related business.   There is a presumption (rebuttable?) that businesses are related if there are a significant number of transfers between them, or share common resources/services.

A TNC will have one or more UT Businesses within it.

For the Report we need to look at the whole Business, not just the element which falls within the countries using UT.

Consolidated accounts

I’m not an accountant, so I shall wave my hands and assume that these can be prepared as required.

Allocation formula

The formula has three inputs:

– Assets
– Labour
– Sales

“Assets” includes all fixed tangible property.  This includes leased assets, to avoid sale-and-leaseback avoidance.  It excludes inventory (and so by extension all other circulating capital?).  It excludes intangibles, because it’s hard to tell where they are and they can move about easily – we look through the intangibles to the assets and staff that create them, and anyway we are more interested in how the IP is used than how it is created.

“Labour” is a combination of payroll costs and headcount; it is the place of work that counts, not the residence of the employer.  They are weighted 50:50.  We include all persons working for the firm, including employees of subcontractors providing labour services – I assume this means staffing agencies rather than all sub-contractors.

“Sales” is the amounts received from selling anything outside the Business, grouped by the residence of the customer rather than the supplier.  There is a reference to the Business needing to have an establishment in the country of the customer, so it can be said that it is doing business there, but this is a much broader term than your normal Article 5 PE under current rules.  The paper is silent about sales to a customer in a country in which you have no establishment.

The three factors need to be weighted in the formula, and there is no conclusion about how this should be done.  It is noted that equal weighting has been usual (though the paper doesn’t specify for whom it is usual).  US states have been tending to go towards 25:25:50, lately (Assets:Labour:Sales).  The EU, on the other hand, has been suggesting 45:45:10.

It is noted that different sectors will have different needs: in particular transport and extractive industries are cited as requiring some careful thought.

The formula itself is not explicitly stated.  I assume that we do the obvious: simply say “Country A has X% of this factor”, take the weighted average of the three percentages, and apply the resulting percentage to the profit of the UT Business in question.  Then we tot up the profits of all Businesses of the TNC in that country, multiply by the headline tax rate, and away we go.


First we had Ptolemy, who worked out that planets revolve around the Earth in perfect circles.  Only his model – being based on a simplified view of the universe that didn’t accord at all well with what actually happened – gave strange results, so he had to add in more and more epicycles to compensate.  The basic model would work if the world were as simple as Ptolemy thought it ought to be, but he was unlucky in the facts he was presented with.

Then we got Copernicus, who worked out that planets revolve around the sun, and Kepler refined this to point out they did so in ellipses rather than circles.  This was a much better model requiring no epicycles, but isn’t quite perfect as it doesn’t really account for the interaction of planets with one another.

So we moved on to proper Newtonian physics, which adds in interplanetary interactions, though it accepts that it’s really complicated and you can’t expect to calculate things precisely so you’re probably going to have to fudge something (but it makes no practical difference).

Unitary Taxation, to me, is Ptolemaic.  Starting with a vision of the universe in which groups are discrete entities (whose activities revolve around tax), it provides a beautiful theoretical model for taxing such entities.  But unfortunately the world is a bit more complicated that it ought to be – groups and companies are not discrete entities, for example – and so to make it work we’d have to start creating epicycles.

The arm’s length principle at the moment seems somewhere between Copernicus and Newton.  You treat companies individually, but you recognise that there is an underlying principle which governs the interactions of any two bodies and use that as a guide when looking at their activities.

BEPS seems to be wanting to move towards a full Newtonian model. It suggests that you have to look at more than the two bodies directly involved when working out the tax position: other companies in the vicinity will have an impact. 

Given that we want to get tax working properly, why should we say that Newtonian mechanics are too hard to use precisely so we should go back to Ptolemy?

Richard Murphy on IP

Richard Murphy has a post on his blog about IP, and why it should be disregarded for tax purposes: http://www.taxresearch.org.uk/Blog/2013/11/13/why-intangibles-should-very-largely-be-ignored-in-international-tax/

I posted a response but as he’d apparently deleted it I thought I’d stick it up here for comment.  (EDIT: actually it was my mistake: he hadn’t deleted it, just hadn’t moderated it in yet.)

His argument basically comes down to:

– It is hard to get an accurate assessment of the value of IP to a business
– Some IP (like his blog) has no value
– Therefore IP doesn’t really add value
– So no profit should be attributed to it (beyond perhaps a nominal amount)
– So for tax purposes it should be disregarded

This would mean that multinationals would be unable to manipulate profit flows, to avoid tax.

I think that by setting the value of IP to nil in all cases you introduce just as many problems – or more, even – than you remove.  You are basically guaranteeing that the value will be wrong.

The examples I gave in my response were (I’m reconstructing here, as the post was deleted, so this is not exactly the same):

– A UK firm builds a brand in the UK, then sets up a subsidiary in France which uses that brand.  If the IP has no value then France gets to tax all the profit of the French company, which is therefore getting profit for nothing.  The UK company will have had a deduction for the costs of building the IP, and so will pay less tax: Murphy’s proposal therefore shifts taxable profit from the UK to France.

– The same firm also licenses the brand to a third party franchisee in France.  There will be a commercial royalty: no firm it its right mnd would allow IP to be used without payment.  If the IP must be ignored for tax purposes we deny a deduction for a cost in France, so the franchisee has a much higher effective tax rate and the UK licensor gets tax-free income.  Alternatively, if we allow a deduction in a third-party situation we introduce a major asymmetry between intra-group and third-party positions, which will mean the tax tail wagging the commercial dog.

I suspect Murphy’s response would be that Unitary Taxation would do away with all these problems, though I note that it would give a very different result: if you include the headcount and costs in the UK of developing the IP, Unitary Taxation would allocate profit to the UK that under the normal rules (but where the IP is considered to be worthless) would stay in France.

The whole suggestion seems somewhat incoherent.  What am I missing?

The case against Starbucks

I’ve been having a brief discussion with Richard Murphy on his blog.  He says that Starbucks was clearly going against the spirit of the law and avoiding tax.

I have asked him for details of this, but he simply asserts that “The case against Starbucks is well known” and “The case was absolutely clear”.

Incidentally, he then deletes my further comments, which I find disappointing given that I am pains to comply with his comment policy, such as by using my real name rather than a pseudonym, and taking care to explain why I reach the conclusions I do.

Unfortunately, although last year I did have a good look at the Starbucks position I didn’t then, and don’t now, understand the detailed case against them.

I know that Murphy and others say they were avoiding tax, but that is a very vague accusation.

I know there were references to coffee beans and royalties, but it is clear that HMRC have reviewed the amounts paid and agreed the rates, so the letter and the spirit of transfer pricing law have been observed.

So far as I am aware there is no permanent establishment issue, as with Google and Amazon – Starbucks was not selling into the UK in a manner taxable elsewhere, but was trading here.

So what is the actual accusation against Starbucks?

The closest I can come to is that Murphy is on record as saying that tax avoidance includes structuring operations so as to follow the law, while not complying with what the law ought to say.  I have serious issues with that contention, but if we assume it to be true for the sake of argument then the only way I can find Starbucks to go against the spirit of the law is to look past the letter of the law, look past the agreed intention of the law, and adduce some deeper spirit which is not apparent.

So when transfer pricing law is considered: the letter is that one should apply the arm’s length principle; the spirit is that a group should not set prices so as to shift profits and erode tax bases; and the deeper spirit is… what?  Do we need to look at tax law in general, rather than transfer pricing, and say that there is some spirit of tax law that says large companies must always pay UK tax?

I really am at a loss.  If anyone can point me to a clear case against Starbucks, such as Murphy assures me exists, I would be very grateful.