The Google Deal: A Response to My Critics

On Saturday evening I published a short post on what Google's tax liability in the UK might look like. You can read it here. It made a number of points, but one in particular has proven controversial.

My post took the revenues that Google Inc, a US entity, reported to its US shareholders as having been made in the UK (£4bn), applied to them the margin that Google Inc makes on its revenues (25%) and to the notional result (£1bn) applied the UK rate of corporation tax 20% to arrive at a notional UK corporation tax liability of £200m. And it compared this figure to the £30m of tax actually paid in the UK, hailed by George Osborne from Davos with more than a little hubris as a "major success".

Critics have said this analysis is wrong. A number of substantially similar examples have been given in the comments section of that post to illustrate just how wrong it is. But let me take the analogous example given by the Financial Secretary to the Treasury in the House of Commons yesterday. He talked of a manufacturer in the UK which makes cars here, ships half of them to Detroit and sells them there. The point I made above, he says, has as a consequence that the sales are taxed in Detroit. And it is wrong, he said, because:

corporation tax is not calculated on the basis of profits attributed to sales in the United Kingdom, but to economic activity and assets located in the United Kingdom.

But this analogy is less to do with the question whether profits are made in Michigan than the conceptually discrete question 'to which territory does the corporation tax system choose to attribute profits'. It does not have as its necessary logical consequence a rebuttal of the argument that profits are made in (in David Gauke's example) Michigan.

We might, in a textbook, predicate a manufacturer in the UK who makes 1,000 widgets here at a cost of 60 each, ships them to Detroit and and sells 300 of them (incurring sales expenses of on average 5) at a cost of 100 per unit. The remainder being sold in the UK. We can hypothesis what his accounts might look like and can construct a sensible argument for attributing different proportions of his profit to the UK and Michigan.

Is his profit in Michigan 300 x 35 (my example above) or 300 x some smaller sum to reflect the fact that we should deem some part of that profit to be attributable to the manufacturing in the UK? An accounting textbook could construct sensible arguments for both.

So I can see the conceptual argument. But what I have more difficulty with its application. Specifically to Google.

Is Google analogous to a car manufacturer?

Not really, it doesn't make stuff and move it and sell it. You might better think of it as an engine that generates money. Is it situated in the US? I don't know, but why should that matter - you could put the machine anywhere. Perhaps the better question is where the machine is built and maintained? But that is to attribute profits to where the costs are rather than where the revenues are from. Why is that logically superior to looking at revenues? Few of us would think of a calculation of profits as starting with costs and applying some uplift. Most - if not all - of us would start with the revenues.

But let's dig a little deeper.

Let's treat the various Google entities as what they are - part of a single economic monolith - and ask what the Google monolith does in the UK?

We know that revenues are generated here.

That is what my notional profit calculation starts from. But the process of generating those revenues is constructed to look like this: finding and warming up sales leads and then passing them to Ireland for signing. If the signing took place in the United Kingdom the tax consequences would be different - and less attractive to Google. To avoid that Google introduces some artifice.

And obviously costs are incurred in generating and warming up those revenues. But not, apparently, costs of a type that entitle the UK to attribute revenue to it. Costs of a different type: costs to which we accept, apparently, that we should apply only a small uplift on which uplift modest tax is charged here.

But generating revenues here is not the only thing Google does here.

Many of the men and women who build and maintain the Google machine live and work here. And they work on building and maintaining the machine. You can see this in the accounts of Google's UK entity. Its activities include the provision of research and development services to Google Inc. Surely these are the 'right' types of costs? The type that entitle us to attribute to the UK a proportion of Google's worldwide revenues?

No.

We apparently accept that they are not that type of cost either. They are also the 'wrong' type of cost. To them, too, we can only apply a modest uplift to generate a modest UK taxable profit.

So. Revenues less margin is wrong, my critics tell me: we should look to where the work is done, to where the cars are made.

But,  in the case of Google, looking at the costs takes us no further. Profits are not taxed where the revenues are earned. Nor on where the costs are incurred. So, where?

To which territory does Google attribute its non US profits?

Not to Ireland. Here's what the Irish Times reported yesterday:

That's a profit on a margin of revenues of less than 1%.

And not to the US either.

Although I cannot corroborate it, it is widely understood that Google Inc has no US tax liability on foreign profits until it repatriates them to the US. And it doesn't.

So although my critics might argue that my calculation of Saturday night fails to respect the textbook norms of international taxation, that argument avails Google only if its affairs do. And they don't.

So I say to my critics, try again. I'm not throwing in the towel yet.