I've just had a conversation with Prof Mike McIntyre of Wayne State University in the USA. He's at the TJN transfer pricing conference in Helsinki.
Our discussion focussed on how easy it is for multinational companies to abuse transfer pricing rules. And it's obvious why. As Mike put it, and I hope I paraphrase correctly:
"The whole logic on which we tax multinationals is wrong. We assume we should compare them with a bunch of unrelated companies. But they aren't. They don't think like that. They don't behave like that. So we try to be force them to be something they're not for tax.
What we should do instead is assume they're what they are - one company with lots of branches.
Mike is right. But of course, as he also knows, just because we should tax them as whole units at the group level we still need to know how to properly allocate that group income to states who can actually collect tax. And for that country-by-country reporting is vital to provide data, and so is a formula based method of allocating profits.
It's not perfect - of course - but as is already clear from the presentations made this morning, the OECD transfer pricing model does not work for lack of data and intellectual and economic incoherence; the Brazilian model works only to the extent that it offers a range of compromises which are accepted as reasonable over-rides to reality and China is struggling to tax whole supply chains - which is great, except they have problems with data capture too, although the method is coherent and recognisably related to the position Mike and I share.
The reality is we're seeking to find a compromise that might mean companies pay something like the right amount of tax in the right place at the right time. The odds are in their favour right now. The challenge is to balance the risks between companies and states. And without reform to existing OECD transfer pricing systems I see now way that will happen.