Let me start with your second question, because I can do that fairly easily, I think.
Let's say you have two related entities, A and B. A provides some kind of value to B, whether it's goods or services or loaned money, or whatever it provides, and B pays A for that value. What transfer pricing does under the arm's-length method is ask whether the amount that B paid A for that value is the correct amount, and it asks that in terms of “What would B have paid A if B and A had not been related?” That's the arm's-length method.
The problem with that is that it is highly factual, particularly when the value has any kind of intangible element to it. Think of a Toyota Motor Corporation in Japan selling Toyota vehicles into a distributor in Canada. The Toyota brand is worth a huge amount of money, but there's not going to be much information on the arm's-length amount that it's worth. So you're at sea, and it really does require some very sophisticated thinking to analyze whether the price that the Toyota distributor pays for those vehicles is the correct price, because a large amount of that price is attributable to the brand and the trademark.
The arm's-length method leads you into a factual swamp where countries of the world try to do the best they can.
A pure formulary method, which I don't necessarily advocate but which a lot of people have, would say forget about the price between Toyota Motor Corporation and the Toyota distributor in Canada; let's look at how much is made from the manufacture of the vehicle to the sale. So you factor out the intercompany transaction, you take the total profit of the sale of the vehicles, and then you divide it up by some formula. Some states of the United States have done that. California uses property, payroll, and sales. So they put in the enumerator of the fraction, property, payroll, and sales in California, and in the denominator property, payroll, and sales worldwide.
There are problems with that formula. I don't want to overstate how good it is, but it's arguably a lot easier to do than to ask yourself what would Toyota, the distributor, have paid for the vehicles if it hadn't been related to the manufacturer? The arm's-length method, the pure arm's-length method, just leads us into these huge never-ending disputes about facts and it does really require an economics Ph.D. to apply this stuff.
Let me go back to your first question. In my country—again, I cannot speak to Canada—we need rules that separate out the havens. I want to remind the committee that the hardest question here is who are we talking about? I understand we're talking about Cayman Islands and the Bahamas and Bermuda, but are we also talking about Ireland and Singapore? Let's face it, if we are, we're into another political world when we start talking about those countries.
In any event, once we decide who we're talking about, I think there ought to be special rules. We shouldn't have the same transfer pricing rules, whatever they are, for both the Cayman Islands and for France. It doesn't make any sense to me. If you have two developed countries where the related parties are situated, you have a very different situation from that where one of those countries is in a tax haven.
So I think countries ought to examine their laws and provide special rules. Transfer pricing is just one area where you could have special rules that are targeted to tax havens. We could do it in the United States if only we would get a little creative—and of course if our political system was functional.
Thank you.