Why Christian Aid's Transfer Mispricing reports are fundamentally flawed
Christian Aid commissioned international trade pricing expert Simon Pak, President of the Trade Research Institute, to analyse EU and US trade data and estimate the amount of capital shifted from non-EU countries into the EU and the US via 'trade mispricing'.
In itself that's not an unreasonable thing to do.
The first problem was that through no fault of his own, Pak was unable to access adequate data for his analysis. The second problem, which Pak himself acknowledges but which tax campaigners don't, is that the methodology Pak applied to the inadequate date was flawed.
What Pak did was this:
- He took detailed trade statistics published by the EU and US. These are widely considered to be robust. I have an issue here.
- He then assigned the types of trade into about 100 different categories, on the reasonable basis that different categories of trade require different levels of skill so will be priced differently. I also have an issue with this because some of the categories he chose were odd, to say the least.
- He then performed a statistical analysis of each category to determine an arm's length range within each one, assuming that anything outside the defined range was mispriced. I have a massive issue with this!
- He then took all "underpriced" imports to and "overpriced" exports from the EU and US (by his pricing definition) and assumed that they represented illicit capital flows out of the developing world and into the EU/US. Again, I have massive issues with this.
- Finally, he applied the relevant country's tax rates on "illicit capital flow" (again, by his definition) to calculate the tax lost to the developing world at $160 billion per annum.
- The reports then assert (and this is nothing to do with Pak) that the flows from the developing world into the EU/US must be routed through tax havens, without providing any evidence to support this.
Let me explain why the flaws are so huge as to make the result meaningless.
First, the trade statistics don't separately identify trade between related companies and trade between unrelated parties. Clearly there's a huge difference. Trade between related companies could be manipulated for tax purposes (because it would benefit the group as a whole to pay less tax). Trade between unrelated parties can't really be manipulated for tax purposes - it can, of course, be manipulated through a fraud perpetrated by the unrelated parties where they privately pocket the difference, but that's not a tax matter - and by definition it's not something that multinationals can possibly get involved in (although I suppose corrupt individuals within multinationals might be - but you can hardly blame the multinationals for that).
Secondly, Pak's statistical analysis took all of the data per trade category and calculated what's called the interquartile range. (Which is, very broadly, the stuff that falls within the middle 50% of each category.) He assumed that anything outside the interquartile range was mispriced. Now that's a load of old crock, because it's a methodology guaranteed to give a result.
Let me demonstrate. Suppose there are 11 boys aged between 3 and 4 in a nursery, and that their heights (in cm) are: 102, 104, 105, 107, 108, 109, 110, 112, 115, 116, 118. The interquartile range is 105 - 115 cm. Now, as it happens all 11 children are within the normal range for 3-4 year old boys. However, using Pak's methodology then four children are "wrong" because they fall outside the interquartile range - the 2 shortest and the 2 tallest. Clearly that's absurd. But it's what Pak did.
Thirdly, 100 categories sounds like a lot, and suggests that they must be very detailed. But two categories provided around 75% of the "mispricing". Unsurprisingly, these were also the categories that included the most diverse products. The biggest "mispriced" category was "nuclear reactors, boilers, mechanical appliances and computers", which according to Pak gave rise to £20.5bn of mispricing. The second biggest was "electrical machinery, sound and TV recorders, telecoms equipment" which was £13.7bn. (The third biggest was "ceramic products" with £2.7 bn - so you can see that there's a massive jump between the top 2 and the rest.) Now, I'm sorry, but how you can possibly compare the pricing of nuclear reactors to the pricing of boilers is utterly beyond me. One is cutting edge, highly regulated technology; the other is boilers. No wonder there were wide ranges either side of the interquartile range! Is there mispricing? No idea - you simply can't tell using this methodology.
Fourthly, Pak only considered "underpriced" imports to and "overpriced" exports from the EU/US. What? He simply ignored "overpriced" imports and "underpriced" exports, both of which would result in flows of funds out of the EU/US! You can't just ignore the half of the data that you say is mispriced and not expect to be called on it.
Fifthly, what was the end result of all this flawed statistical work from Pak? Not what you'd expect form the headlines. Fully 80% of the "mispriced" flows into the EU/US came from countries classes as high or upper middle income countries. Not from the third world at all. Just over 3% flowed from low income countries. And in fact the biggest "illicit" flows into the EU came from the US, Japan, China and Turkey. The biggest "illicit" flows into the US came from Malaysia, Japan, China, Canada and Mexico. There's not a single third world country in the top 10 for either. So all that Pak's results demonstrate is that rich nations are largely trading with other rich nations. It certainly doesn't prove that greedy multinationals are systematically stripping profits out of the third world - quite the opposite. How can you strip out profits from the third world if you're not doing business there in the first place?
Finally, as a general rule taxes are much higher in the developed world than in the developing world. (That's true even allowing for the fact that headline corporate tax rates are broadly similar, or even lower, in the developed world. The reason is that the developed world taxes a much wider base of profits then the developing world.) So it makes absolutely no sense for a multinational to shift profits from low tax countries to high tax countries, if the object of the exercise is saving tax!
So to summarise:
- The reports start with flawed data
- They then apply flawed methodology to the flawed data
- All the end results of the analysis can possibly demonstrate is that largely multinationals aren't trading with the third world at all and therefore can't be systematically stripping profits out
- In any case, it doesn't make sense from a tax perspective to move profits from the third world to the first world anyhow.
- there's no evidence whatsoever, anywhere in the reports, that tax havens are involved in any tax abuse (it's just taken as a given, which annoys me greatly).